From shaojieb at andrew.cmu.edu  Tue Feb  2 07:49:45 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Tue, 2 Feb 2021 07:49:45 -0500
Subject: [AI Seminar] AI Seminar sponsored by Fortive -- Feb 09 (Zoom) --
 Michael Auli (FAIR) -- Self-supervised Learning of Speech Representations
 with wav2vec
Message-ID: <CAJA0PPAZDdnqk_WeYf-O06ftRaQY=sn8ZUi=kbTuv0uuf5o4_w@mail.gmail.com>

Dear all,

Happy new year, and welcome back!

We look forward to seeing you *next Tuesday (2/9)* from 12:00-1:00 PM (U.S.
Eastern time) for the first talk this semester of our *CMU AI seminar*,
sponsored by Fortive <https://careers.fortive.com/>.

To learn more about the seminar series, subscribe to its mailing list, or
see the future schedule, please visit the seminar website
<http://www.cs.cmu.edu/~aiseminar/>. <http://www.cs.cmu.edu/~aiseminar/>

On 2/9, Michael Auli <https://michaelauli.github.io/> (Facebook AI
Research) will be giving a talk on "*Self-supervised Learning of Speech
Representations with wav2vec*."

*Title*: Self-supervised Learning of Speech Representations with wav2vec

*Talk Abstract*: Self-supervised learning has been a key driver of progress
in natural language processing and increasingly in computer vision. In this
talk, I will give an overview of the wav2vec line of work which explores
algorithms to learn good representations of speech audio solely from
unlabeled data. The resulting models can be fine-tuned for a specific task
using labeled data and enable speech recognition models with just 10
minutes of labeled speech audio by leveraging a large amount of unlabeled
speech. Our latest work, wav2vec 2.0 learns a vocabulary of speech units
obtained by quantizing the latent representation of the speech signal and
by solving a contrastive task defined over the quantization. We also
explored multilingual pre-training and recently released a model trained on
53 different languages.

*Speaker Bio*: Michael Auli is a scientist at Facebook AI Research in Menlo
Park, California. During his PhD, he worked on natural language processing
and parsing at the University of Edinburgh where he was advised by Adam
Lopez and Philipp Koehn. While at Microsoft Research, he did some of the
early work on neural machine translation and neural dialogue models. After
this, he led the team which developed convolutional sequence to sequence
models that were the first models to outperform recurrent neural networks
for neural machine translation. Currently, Michael works on semi-supervised
and self-supervised learning applied to natural language processing and
speech recognition. He led the teams which ranked first in several tasks of
the WMT news translation task in 2018 and 2019.

*Zoom Link*:
https://cmu.zoom.us/j/91735176241?pwd=WmhzdWdJT2IxN2Y4ZGt6WnduKzRDUT09
<https://cmu.zoom.us/j/91248740598?pwd=V1M2L09xRlR6V3gvK2VVNERzOHRuQT09>


Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210202/92162653/attachment.html>

From shaojieb at andrew.cmu.edu  Mon Feb  8 12:08:18 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Mon, 8 Feb 2021 12:08:18 -0500
Subject: [CMU AI Seminar] Fwd: AI Seminar sponsored by Fortive -- Feb 09
 (Zoom) -- Michael Auli (FAIR) -- Self-supervised Learning of Speech
 Representations with wav2vec
In-Reply-To: <CAJA0PPAZDdnqk_WeYf-O06ftRaQY=sn8ZUi=kbTuv0uuf5o4_w@mail.gmail.com>
References: <CAJA0PPAZDdnqk_WeYf-O06ftRaQY=sn8ZUi=kbTuv0uuf5o4_w@mail.gmail.com>
Message-ID: <CAJA0PPDi-viFaaeP2bofK7RRB8-wJGew4EqEaJCZ3QnHkXNEjA@mail.gmail.com>

Hi all,

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow 12pm-1pm:
https://cmu.zoom.us/j/91735176241?pwd=WmhzdWdJT2IxN2Y4ZGt6WnduKzRDUT09.

Michael Auli (FAIR) will be talking about self-supervised learning of
speech representations (see below).

Thanks,
Shaojie

---------- Forwarded message ---------
From: Shaojie Bai <shaojieb at andrew.cmu.edu>
Date: Tue, Feb 2, 2021 at 7:49 AM
Subject: AI Seminar sponsored by Fortive -- Feb 09 (Zoom) -- Michael Auli
(FAIR) -- Self-supervised Learning of Speech Representations with wav2vec
To: <ai-seminar-announce at cs.cmu.edu>
Cc: <lti-students at cs.cmu.edu>, <cs-students at cs.cmu.edu>, <
ml-students at cs.cmu.edu>, <robotics-students at cs.cmu.edu>, <
ece-students at ece.cmu.edu>


Dear all,

Happy new year, and welcome back!

We look forward to seeing you *next Tuesday (2/9)* from 12:00-1:00 PM (U.S.
Eastern time) for the first talk this semester of our *CMU AI seminar*,
sponsored by Fortive <https://careers.fortive.com/>.

To learn more about the seminar series, subscribe to its mailing list, or
see the future schedule, please visit the seminar website
<http://www.cs.cmu.edu/~aiseminar/>. <http://www.cs.cmu.edu/~aiseminar/>

On 2/9, Michael Auli <https://michaelauli.github.io/> (Facebook AI
Research) will be giving a talk on "*Self-supervised Learning of Speech
Representations with wav2vec*."

*Title*: Self-supervised Learning of Speech Representations with wav2vec

*Talk Abstract*: Self-supervised learning has been a key driver of progress
in natural language processing and increasingly in computer vision. In this
talk, I will give an overview of the wav2vec line of work which explores
algorithms to learn good representations of speech audio solely from
unlabeled data. The resulting models can be fine-tuned for a specific task
using labeled data and enable speech recognition models with just 10
minutes of labeled speech audio by leveraging a large amount of unlabeled
speech. Our latest work, wav2vec 2.0 learns a vocabulary of speech units
obtained by quantizing the latent representation of the speech signal and
by solving a contrastive task defined over the quantization. We also
explored multilingual pre-training and recently released a model trained on
53 different languages.

*Speaker Bio*: Michael Auli is a scientist at Facebook AI Research in Menlo
Park, California. During his PhD, he worked on natural language processing
and parsing at the University of Edinburgh where he was advised by Adam
Lopez and Philipp Koehn. While at Microsoft Research, he did some of the
early work on neural machine translation and neural dialogue models. After
this, he led the team which developed convolutional sequence to sequence
models that were the first models to outperform recurrent neural networks
for neural machine translation. Currently, Michael works on semi-supervised
and self-supervised learning applied to natural language processing and
speech recognition. He led the teams which ranked first in several tasks of
the WMT news translation task in 2018 and 2019.

*Zoom Link*:
https://cmu.zoom.us/j/91735176241?pwd=WmhzdWdJT2IxN2Y4ZGt6WnduKzRDUT09


Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210208/3ed98f95/attachment.html>

From shaojieb at andrew.cmu.edu  Tue Feb  9 13:37:00 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Tue, 9 Feb 2021 13:37:00 -0500
Subject: [CMU AI Seminar] AI Seminar sponsored by Fortive -- Feb 16 (Zoom)
 -- Will Grathwohl (U of Toronto) -- Using and Abusing Gradients for
 Discrete Sampling and Energy-Based Models
Message-ID: <CAJA0PPDA3Ro+tRTfZQ3M4EyWVbdn2dxAc76_UC49id2xbPmEmA@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (2/16)* from 12:00-1:00 PM
(U.S. Eastern time) for the next talk of our *CMU AI seminar*, sponsored by
Fortive <https://careers.fortive.com/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
<http://www.cs.cmu.edu/~aiseminar/>

On 2/16, *Will Grathwohl <http://www.cs.toronto.edu/~wgrathwohl/>* (University
of Toronto) will be giving a talk on "*Using and Abusing Gradients for
Discrete Sampling and Energy-Based Models*."

*Title*: Using and Abusing Gradients for Discrete Sampling and Energy-Based
Models

*Talk Abstract*: Deep energy-based models have quickly become a popular and
successful approach for generative modeling in high dimensions. The success
of these models can mainly be attributed to improvements in MCMC sampling
such as Langevin Dynamics and training with large persistent Markov chains.
Because of the reliance on gradient-based sampling, these models have been
successful in modeling continuous data. As it stands, the same solution
cannot be applied when dealing with discrete data. In this work, we propose
a general and automatic approximate sampling strategy for probabilistic
models with discrete variables. Our approach uses gradients of the
likelihood function with respect to discrete assignments to propose
Metropolis-Hastings updates. These updates can be incorporated into larger
Markov chain Monte Carlo or learning schemes. We show theoretically and
empirically that this simple approach outperforms generic samplers in a
number of difficult settings including Ising Models, Potts Models,
Restricted Boltzmann Machines, and Factorial Hidden Markov Models -- even
outperforming some samplers which exploit known structure in these
distributions. We also show that our improved sampler enables the training
of deep energy-based models on high dimensional discrete data which
outperforms variational auto-encoders and previous energy-based models.

*Speaker Bio*: Will Grathwohl is a PhD student at the University of Toronto
supervised by Richard Zemel and David Duvenaud. His work mainly focuses on
generative models and their applications to downstream discriminative
tasks. His work has covered variational inference, normalizing flows and
now focuses mainly on energy-based models. Will is currently a student
researcher on the Google Brain team in Toronto. Prior to graduate school,
Will worked on machine learning applications in silicon valley and did his
undergraduate degree in mathematics at the Massachusetts Institute of
Technology.

*Zoom Link*:
https://cmu.zoom.us/j/91853143684?pwd=UDNLNWpRcEs2WUx4S21UZ3d2RHV2dz09


Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210209/96229bec/attachment.html>

From shaojieb at andrew.cmu.edu  Mon Feb 15 11:55:44 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Mon, 15 Feb 2021 11:55:44 -0500
Subject: [CMU AI Seminar] Fwd: AI Seminar sponsored by Fortive -- Feb 16
 (Zoom) -- Will Grathwohl (U of Toronto) -- Using and Abusing Gradients for
 Discrete Sampling and Energy-Based Models
In-Reply-To: <CAJA0PPDA3Ro+tRTfZQ3M4EyWVbdn2dxAc76_UC49id2xbPmEmA@mail.gmail.com>
References: <CAJA0PPDA3Ro+tRTfZQ3M4EyWVbdn2dxAc76_UC49id2xbPmEmA@mail.gmail.com>
Message-ID: <CAJA0PPBdns1UEHc8hy2WtuyXq6yR9X3+soPBhaXwrouuq8hy5Q@mail.gmail.com>

Hi all,

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow 12pm-1pm:
https://cmu.zoom.us/j/91853143684?pwd=UDNLNWpRcEs2WUx4S21UZ3d2RHV2dz09.

Will Grathwohl (University of Toronto) will be talking about discrete
sampling and deep energy-based models (see below).

Thanks,
Shaojie

---------- Forwarded message ---------
From: Shaojie Bai <shaojieb at andrew.cmu.edu>
Date: Tue, Feb 9, 2021 at 1:37 PM
Subject: AI Seminar sponsored by Fortive -- Feb 16 (Zoom) -- Will Grathwohl
(U of Toronto) -- Using and Abusing Gradients for Discrete Sampling and
Energy-Based Models
To: <ai-seminar-announce at cs.cmu.edu>


Dear all,

We look forward to seeing you *next Tuesday (2/16)* from 12:00-1:00 PM
(U.S. Eastern time) for the next talk of our *CMU AI seminar*, sponsored by
Fortive <https://careers.fortive.com/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
<http://www.cs.cmu.edu/~aiseminar/>

On 2/16, *Will Grathwohl <http://www.cs.toronto.edu/~wgrathwohl/>* (University
of Toronto) will be giving a talk on "*Using and Abusing Gradients for
Discrete Sampling and Energy-Based Models*."

*Title*: Using and Abusing Gradients for Discrete Sampling and Energy-Based
Models

*Talk Abstract*: Deep energy-based models have quickly become a popular and
successful approach for generative modeling in high dimensions. The success
of these models can mainly be attributed to improvements in MCMC sampling
such as Langevin Dynamics and training with large persistent Markov chains.
Because of the reliance on gradient-based sampling, these models have been
successful in modeling continuous data. As it stands, the same solution
cannot be applied when dealing with discrete data. In this work, we propose
a general and automatic approximate sampling strategy for probabilistic
models with discrete variables. Our approach uses gradients of the
likelihood function with respect to discrete assignments to propose
Metropolis-Hastings updates. These updates can be incorporated into larger
Markov chain Monte Carlo or learning schemes. We show theoretically and
empirically that this simple approach outperforms generic samplers in a
number of difficult settings including Ising Models, Potts Models,
Restricted Boltzmann Machines, and Factorial Hidden Markov Models -- even
outperforming some samplers which exploit known structure in these
distributions. We also show that our improved sampler enables the training
of deep energy-based models on high dimensional discrete data which
outperforms variational auto-encoders and previous energy-based models.

*Speaker Bio*: Will Grathwohl is a PhD student at the University of Toronto
supervised by Richard Zemel and David Duvenaud. His work mainly focuses on
generative models and their applications to downstream discriminative
tasks. His work has covered variational inference, normalizing flows and
now focuses mainly on energy-based models. Will is currently a student
researcher on the Google Brain team in Toronto. Prior to graduate school,
Will worked on machine learning applications in silicon valley and did his
undergraduate degree in mathematics at the Massachusetts Institute of
Technology.

*Zoom Link*:
https://cmu.zoom.us/j/91853143684?pwd=UDNLNWpRcEs2WUx4S21UZ3d2RHV2dz09


Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210215/e88ca9e6/attachment.html>

From shaojieb at andrew.cmu.edu  Tue Feb 23 12:15:17 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Tue, 23 Feb 2021 12:15:17 -0500
Subject: [CMU AI Seminar] Mar 2 (Zoom) -- Vaishnavh Nagarajan (CMU) --
 Understanding the Failure Modes of Out-of-Distribution Generalization -- AI
 Seminar sponsored by Fortive
Message-ID: <CAJA0PPDWUU5CXQ-wtoXVJx0RP6=7JpQZvyNFgMq3WzLR4BRw4w@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (3/2)* from 12:00-1:00 PM (U.S.
Eastern time) for the next talk of our *CMU AI seminar*, sponsored by
Fortive <https://careers.fortive.com/>.

To learn more about the seminar series, see the future schedule, or add the
seminar calendar to your own, please visit the seminar website
<http://www.cs.cmu.edu/~aiseminar/>. <http://www.cs.cmu.edu/~aiseminar/>

On 3/2, *Vaishnavh Nagarajan* (CMU Computer Science Department) will be
giving a talk on "*Understanding the Failure Modes of Out-of-Distribution
Generalization*."

*Title*: Understanding the Failure Modes of Out-of-Distribution
Generalization

*Talk Abstract*: Classifiers often rely on features like the background
that may be spuriously correlated with the label. In practice, this results
in poor test-time accuracy as the classifier may be deployed in an
environment where these spurious correlations no longer hold. While many
algorithms have been developed to heuristically tackle this challenge of
out-of-distribution generalization, in this work, we take a step back to
ask: why do classifiers rely on spurious correlations in the first place?
While the answer to this might seem straightforward, I'll begin by
explaining why existing theoretical models of spurious correlations do not
capture the fundamental reasons behind why classifiers rely on spurious
correlations. I'll then propose an alternative theoretical model which
helps uncover those fundamental reasons. In particular, by theoretically
studying linear classifiers in this theoretical model, we'll look at two
failure modes: one that is "geometric" in nature another that is
"statistical" in nature. These modes shed insight to the exact biases in
gradient descent, and the exact properties of real-world data that
incentivize classifiers to use spurious correlations. Finally, I'll discuss
experiments on neural networks that validate these insights in more
practical scenarios. Hopefully, with the knowledge of these failure modes,
algorithm designers can be better informed about how to fix these failure
modes, and OoD research can be built upon a more rigorous foundation.

*Speaker Bio*: Vaishnavh Nagarajan is a final year Computer Science PhD
student at Carnegie Mellon University (CMU), advised by Zico Kolter.
Vaishnavh is broadly interested in the theoretical foundations of machine
learning, involving problems in the intersection of learning theory and
optimization. He is particularly interested in theoretically understanding
when and why modern machine learning algorithms work (or do not work) in
practice. His work has received an Outstanding New Directions Paper Award
at NeurIPS'19, an oral presentation at NeurIPS'17 and three workshop
spotlight talks. Prior to CMU, Vaishnavh completed his undergraduate
studies in Computer Science and Engineering in the Indian Institute of
Technology, Madras.

*Zoom Link*:
https://cmu.zoom.us/j/94477694788?pwd=WmdvNTlsUW5oRHl1dFRDbzkrVmVNdz09


Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210223/370828a6/attachment.html>

From shaojieb at andrew.cmu.edu  Mon Mar  1 11:43:32 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Mon, 1 Mar 2021 11:43:32 -0500
Subject: [CMU AI Seminar] Fwd: Mar 2 (Zoom) -- Vaishnavh Nagarajan (CMU) --
 Understanding the Failure Modes of Out-of-Distribution Generalization -- AI
 Seminar sponsored by Fortive
In-Reply-To: <CAJA0PPDWUU5CXQ-wtoXVJx0RP6=7JpQZvyNFgMq3WzLR4BRw4w@mail.gmail.com>
References: <CAJA0PPDWUU5CXQ-wtoXVJx0RP6=7JpQZvyNFgMq3WzLR4BRw4w@mail.gmail.com>
Message-ID: <CAJA0PPD4q4oihaz+8-OnC7GYfNyC2TuFr_jCdcq2kAf2Lo=7nA@mail.gmail.com>

Hi all,

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow 12pm-1pm:
https://cmu.zoom.us/j/94477694788?pwd=WmdvNTlsUW5oRHl1dFRDbzkrVmVNdz09.

Vaishnavh Nagarajan (CMU) will be talking about the theories and intuitions
behind the failure modes of out-of-distribution generalization (see below).

Thanks,
Shaojie

---------- Forwarded message ---------
From: Shaojie Bai <shaojieb at andrew.cmu.edu>
Date: Tue, Feb 23, 2021 at 12:15 PM
Subject: Mar 2 (Zoom) -- Vaishnavh Nagarajan (CMU) -- Understanding the
Failure Modes of Out-of-Distribution Generalization -- AI Seminar sponsored
by Fortive
To: <ai-seminar-announce at cs.cmu.edu>


Dear all,

We look forward to seeing you *next Tuesday (3/2)* from 12:00-1:00 PM (U.S.
Eastern time) for the next talk of our *CMU AI seminar*, sponsored by
Fortive <https://careers.fortive.com/>.

To learn more about the seminar series, see the future schedule, or add the
seminar calendar to your own, please visit the seminar website
<http://www.cs.cmu.edu/~aiseminar/>. <http://www.cs.cmu.edu/~aiseminar/>

On 3/2, *Vaishnavh Nagarajan* (CMU Computer Science Department) will be
giving a talk on "*Understanding the Failure Modes of Out-of-Distribution
Generalization*."

*Title*: Understanding the Failure Modes of Out-of-Distribution
Generalization

*Talk Abstract*: Classifiers often rely on features like the background
that may be spuriously correlated with the label. In practice, this results
in poor test-time accuracy as the classifier may be deployed in an
environment where these spurious correlations no longer hold. While many
algorithms have been developed to heuristically tackle this challenge of
out-of-distribution generalization, in this work, we take a step back to
ask: why do classifiers rely on spurious correlations in the first place?
While the answer to this might seem straightforward, I'll begin by
explaining why existing theoretical models of spurious correlations do not
capture the fundamental reasons behind why classifiers rely on spurious
correlations. I'll then propose an alternative theoretical model which
helps uncover those fundamental reasons. In particular, by theoretically
studying linear classifiers in this theoretical model, we'll look at two
failure modes: one that is "geometric" in nature another that is
"statistical" in nature. These modes shed insight to the exact biases in
gradient descent, and the exact properties of real-world data that
incentivize classifiers to use spurious correlations. Finally, I'll discuss
experiments on neural networks that validate these insights in more
practical scenarios. Hopefully, with the knowledge of these failure modes,
algorithm designers can be better informed about how to fix these failure
modes, and OoD research can be built upon a more rigorous foundation.

*Speaker Bio*: Vaishnavh Nagarajan is a final year Computer Science PhD
student at Carnegie Mellon University (CMU), advised by Zico Kolter.
Vaishnavh is broadly interested in the theoretical foundations of machine
learning, involving problems in the intersection of learning theory and
optimization. He is particularly interested in theoretically understanding
when and why modern machine learning algorithms work (or do not work) in
practice. His work has received an Outstanding New Directions Paper Award
at NeurIPS'19, an oral presentation at NeurIPS'17 and three workshop
spotlight talks. Prior to CMU, Vaishnavh completed his undergraduate
studies in Computer Science and Engineering in the Indian Institute of
Technology, Madras.

*Zoom Link*:
https://cmu.zoom.us/j/94477694788?pwd=WmdvNTlsUW5oRHl1dFRDbzkrVmVNdz09


Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210301/9eda43e8/attachment.html>

From shaojieb at andrew.cmu.edu  Tue Mar  2 13:10:22 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Tue, 2 Mar 2021 13:10:22 -0500
Subject: [CMU AI Seminar] Mar 9 (Zoom) -- Kayvon Fatahalian (Stanford) --
 Keeping the Domain Expert in the Loop: Ideas to Models in Hours, Not Weeks --
 AI Seminar sponsored by Fortive
Message-ID: <CAJA0PPDZSAcVtjMqf14gUG-sJpK+=Kgoo0HiLFXjb3U4hFygbw@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (3/9)* from 12:00-1:00 PM (U.S.
Eastern time) for the next talk of our *CMU AI seminar*, sponsored by
Fortive <https://careers.fortive.com/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
<http://www.cs.cmu.edu/~aiseminar/>

On 3/9, *Kayvon Fatahalian <http://graphics.stanford.edu/~kayvonf/>* (Stanford
University) will be giving a talk on "*Keeping the Domain Expert in the
Loop: Ideas to Models in Hours, Not Weeks*."

*Title*: Keeping the Domain Expert in the Loop: Ideas to Models in Hours,
Not Weeks

*Talk Abstract*: My students and I often find ourselves as "subject matter
experts" needing to create models for use in big data computer graphics and
video analysis applications. Yet it is frustrating that a capable grad
student, armed with a large unlabeled image/video collection, a palette of
modern pre-trained models, and an idea of what novel object or event they
want to detect, still requires days-to-weeks to create good models for
their task. In this talk, I will discuss challenges we've faced carrying
out the iterative process of data curation, model training, and model
validation for the specific case of rare events and categories in image and
video collections (such as professional broadcast sports and cable TV). Our
ultimate goal (not yet achieved) is to create training techniques and data
selection interfaces that enable interactive, grad-student-in-the-loop
workflows where the expert human is working concurrently with massive
amounts of parallel processing to interactively and continuously perform
cycles of data acquisition, training, and validation.

*Speaker Bio*: Kayvon Fatahalian is an Assistant Professor in the Computer
Science Department at Stanford University. His lab works on visual
computing systems projects, including high-performance rendering for RL,
large-scale video analytics, programming systems for video data mining, and
compilation techniques for optimizing image processing pipelines. In all
these efforts, the goal is to enable rapid development of applications that
involve image and video processing at scale.

*Zoom Link*:
https://cmu.zoom.us/j/92196990617?pwd=Zk1vZkhzbTkwTE1nNzcyMm5JTFRpUT09


Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210302/4fd89725/attachment.html>

From shaojieb at andrew.cmu.edu  Mon Mar  8 12:11:36 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Mon, 8 Mar 2021 12:11:36 -0500
Subject: [CMU AI Seminar] Mar 9 (Zoom) -- Kayvon Fatahalian (Stanford) --
 Keeping the Domain Expert in the Loop: Ideas to Models in Hours,
 Not Weeks -- AI Seminar sponsored by Fortive
In-Reply-To: <CAJA0PPDZSAcVtjMqf14gUG-sJpK+=Kgoo0HiLFXjb3U4hFygbw@mail.gmail.com>
References: <CAJA0PPDZSAcVtjMqf14gUG-sJpK+=Kgoo0HiLFXjb3U4hFygbw@mail.gmail.com>
Message-ID: <CAJA0PPB-0VmPFf2K_sYnczNn9zJq1m7L__pRRxLvchUJaoyxUw@mail.gmail.com>

Hi all,

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow 12pm-1pm:
https://cmu.zoom.us/j/92196990617?pwd=Zk1vZkhzbTkwTE1nNzcyMm5JTFRpUT09.

*Kayvon Fatahalian* (Stanford University) will be talking about how we
could come up with great model ideas and training workflows on
images/videos quickly in practice (see below).

Thanks,
Shaojie

On Tue, Mar 2, 2021 at 1:10 PM Shaojie Bai <shaojieb at andrew.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (3/9)* from 12:00-1:00 PM
> (U.S. Eastern time) for the next talk of our *CMU AI seminar*, sponsored
> by Fortive <https://careers.fortive.com/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
> <http://www.cs.cmu.edu/~aiseminar/>
>
> On 3/9, *Kayvon Fatahalian <http://graphics.stanford.edu/~kayvonf/>* (Stanford
> University) will be giving a talk on "*Keeping the Domain Expert in the
> Loop: Ideas to Models in Hours, Not Weeks*."
>
> *Title*: Keeping the Domain Expert in the Loop: Ideas to Models in Hours,
> Not Weeks
>
> *Talk Abstract*: My students and I often find ourselves as "subject
> matter experts" needing to create models for use in big data computer
> graphics and video analysis applications. Yet it is frustrating that a
> capable grad student, armed with a large unlabeled image/video collection,
> a palette of modern pre-trained models, and an idea of what novel object or
> event they want to detect, still requires days-to-weeks to create good
> models for their task. In this talk, I will discuss challenges we've faced
> carrying out the iterative process of data curation, model training, and
> model validation for the specific case of rare events and categories in
> image and video collections (such as professional broadcast sports and
> cable TV). Our ultimate goal (not yet achieved) is to create training
> techniques and data selection interfaces that enable interactive,
> grad-student-in-the-loop workflows where the expert human is working
> concurrently with massive amounts of parallel processing to interactively
> and continuously perform cycles of data acquisition, training, and
> validation.
>
> *Speaker Bio*: Kayvon Fatahalian is an Assistant Professor in the
> Computer Science Department at Stanford University. His lab works on visual
> computing systems projects, including high-performance rendering for RL,
> large-scale video analytics, programming systems for video data mining, and
> compilation techniques for optimizing image processing pipelines. In all
> these efforts, the goal is to enable rapid development of applications that
> involve image and video processing at scale.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/92196990617?pwd=Zk1vZkhzbTkwTE1nNzcyMm5JTFRpUT09
>
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210308/7d2a376a/attachment.html>

From shaojieb at andrew.cmu.edu  Tue Mar  9 13:13:21 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Tue, 9 Mar 2021 13:13:21 -0500
Subject: [CMU AI Seminar] Mar 16 (Zoom) -- Raia Hadsell (DeepMind) -- Scalable
 Robot Learning in Rich Environments -- AI Seminar sponsored by Fortive
Message-ID: <CAJA0PPC-7HT6ZYLUX4=OdFfDZ0+pMwCcG+X-DBFe1gYa5EXOnQ@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (3/16)* from 12:00-1:00 PM
(U.S. Eastern time) for the next talk of our *CMU AI seminar*, sponsored by
Fortive <https://careers.fortive.com/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
<http://www.cs.cmu.edu/~aiseminar/>

On 3/16, *Raia Hadsell <http://raiahadsell.com/index.html>* (DeepMind) will
be giving a talk on "*Scalable Robot Learning in Rich Environments*."

*Title*: Scalable Robot Learning in Rich Environments

*Talk Abstract*: As modern machine learning methods push towards
breakthroughs in controlling physical systems, games and simple physical
simulations are often used as the main benchmark domains. As the field
matures, it is important to develop more sophisticated learning systems
with the aim of solving more complex real-world tasks, but problems like
catastrophic forgetting and data efficiency remain critical, particularly
for robotic domains. This talk will cover some of the challenges that exist
for learning from interactions in more complex, constrained, and real-world
settings, and some promising new approaches that have emerged.

*Speaker Bio*: Raia Hadsell is the Director of Robotics at DeepMind. Dr.
Hadsell joined DeepMind in 2014 to pursue new solutions for artificial
general intelligence. Her research focuses on the challenge of continual
learning for AI agents and robots, and she has proposed neural approaches
such as policy distillation, progressive nets, and elastic weight
consolidation to solve the problem of catastrophic forgetting. Dr. Hadsell
is on the executive boards of ICLR (International Conference on Learning
Representations), WiML (Women in Machine Learning), and CoRL (Conference on
Robot Learning). She is a fellow of the European Lab on Learning Systems
(ELLIS), a founding organizer of NAISys (Neuroscience for AI Systems), and
serves as a CIFAR advisor.

*Zoom Link*:
https://cmu.zoom.us/j/93418102649?pwd=TTd4dElxWnBOZHJ5QndUNVBWUjZCZz09


Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210309/c072fd57/attachment.html>

From shaojieb at andrew.cmu.edu  Mon Mar 15 13:00:30 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Mon, 15 Mar 2021 13:00:30 -0400
Subject: [CMU AI Seminar] Mar 16 (Zoom) -- Raia Hadsell (DeepMind) --
 Scalable Robot Learning in Rich Environments -- AI Seminar sponsored by
 Fortive
In-Reply-To: <CAJA0PPC-7HT6ZYLUX4=OdFfDZ0+pMwCcG+X-DBFe1gYa5EXOnQ@mail.gmail.com>
References: <CAJA0PPC-7HT6ZYLUX4=OdFfDZ0+pMwCcG+X-DBFe1gYa5EXOnQ@mail.gmail.com>
Message-ID: <CAJA0PPBq=-J8-+3cxiko=LT8Lk=o2=CRzgkLcsaJ_7+K=uUj+Q@mail.gmail.com>

Hi all,

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow 12pm-1pm:
https://cmu.zoom.us/j/93418102649?pwd=TTd4dElxWnBOZHJ5QndUNVBWUjZCZz09.

Raia Hadsell (DeepMind) will be talking about some recent challenges
involved in scalable robot learning (see below).

Thanks,
Shaojie

On Tue, Mar 9, 2021 at 1:13 PM Shaojie Bai <shaojieb at andrew.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (3/16)* from 12:00-1:00 PM
> (U.S. Eastern time) for the next talk of our *CMU AI seminar*, sponsored
> by Fortive <https://careers.fortive.com/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
> <http://www.cs.cmu.edu/~aiseminar/>
>
> On 3/16, *Raia Hadsell <http://raiahadsell.com/index.html>* (DeepMind)
> will be giving a talk on "*Scalable Robot Learning in Rich Environments*."
>
> *Title*: Scalable Robot Learning in Rich Environments
>
> *Talk Abstract*: As modern machine learning methods push towards
> breakthroughs in controlling physical systems, games and simple physical
> simulations are often used as the main benchmark domains. As the field
> matures, it is important to develop more sophisticated learning systems
> with the aim of solving more complex real-world tasks, but problems like
> catastrophic forgetting and data efficiency remain critical, particularly
> for robotic domains. This talk will cover some of the challenges that exist
> for learning from interactions in more complex, constrained, and real-world
> settings, and some promising new approaches that have emerged.
>
> *Speaker Bio*: Raia Hadsell is the Director of Robotics at DeepMind. Dr.
> Hadsell joined DeepMind in 2014 to pursue new solutions for artificial
> general intelligence. Her research focuses on the challenge of continual
> learning for AI agents and robots, and she has proposed neural approaches
> such as policy distillation, progressive nets, and elastic weight
> consolidation to solve the problem of catastrophic forgetting. Dr. Hadsell
> is on the executive boards of ICLR (International Conference on Learning
> Representations), WiML (Women in Machine Learning), and CoRL (Conference on
> Robot Learning). She is a fellow of the European Lab on Learning Systems
> (ELLIS), a founding organizer of NAISys (Neuroscience for AI Systems), and
> serves as a CIFAR advisor.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/93418102649?pwd=TTd4dElxWnBOZHJ5QndUNVBWUjZCZz09
>
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210315/8ac33cf5/attachment.html>

From shaojieb at andrew.cmu.edu  Tue Mar 16 22:24:17 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Tue, 16 Mar 2021 22:24:17 -0400
Subject: [CMU AI Seminar] Mar 23 (Zoom) -- Le Song (Georgia Tech/MBZUAI) --
 Understanding Deep Architectures with Reasoning Layer -- AI Seminar sponsored
 by Fortive
Message-ID: <CAJA0PPD4brVyRd1B+Naqaj6+eYkjkXnk=YNJqBna6cMvjU5RWw@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (3/23)* from 12:00-1:00 PM
(U.S. Eastern time) for the next talk of our *CMU AI seminar*, sponsored by
Fortive <https://careers.fortive.com/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
<http://www.cs.cmu.edu/~aiseminar/>

On 3/23, *Le Song* (Georgia Tech / MBZUAI) will be giving a talk on
"*Understanding
Deep Architectures with Reasoning Layer*."

*Title*: Understanding Deep Architectures with Reasoning Layer

*Talk Abstract*: Recently, there has been a surge of interest in combining
deep learning models with reasoning in order to handle more sophisticated
learning tasks. In many cases, a reasoning task can be solved by an
iterative algorithm. This algorithm is often unrolled, and used as a
specialized layer in the deep architecture, which can be trained end-to-end
with other neural components. Although such hybrid deep architectures have
led to many empirical successes, the theoretical foundation of such
architectures, especially the interplay between algorithm layers and other
neural layers, remains largely unexplored. In this paper, we take an
initial step towards an understanding of such hybrid deep architectures by
showing that properties of the algorithm layers, such as convergence,
stability and sensitivity, are intimately related to the approximation and
generalization abilities of the end-to-end model. Furthermore, our analysis
matches closely our experimental observations under various conditions,
suggesting that our theory can provide useful guidelines for designing deep
architectures with reasoning layers.

*Speaker Bio*: Le Song is a Professor and the Deputy Chair of the Machine
Learning Department, Mohamed bin Sayed University of AI, UAE. He was an
Associate Professor in the Department of Computational Science and
Engineering, College of Computing, Georgia Institute of Technology, worked
as a research scientist at Google, and did his post-doc in Carnegie Mellon
University. His principal research area is machine learning, especially
kernel methods, deep learning, and probabilistic graphical models. He is
the recipient of many best paper awards at major machine learning
conferences, such as NeurIPS, ICML and AISTATS, and the NSF CAREER Award.
He has also served as the area chair or senior program committee for many
leading machine learning and AI conferences such as NeurIPS, ICML, ICLR,
AAAI and IJCAI, and the action editor for JMLR and IEEE TPAMI. He is also a
board member of the International Conference for Machine Learning.

*Zoom Link*:
https://cmu.zoom.us/j/94092736800?pwd=OWlCRlV6RERtRHByWWpkZEl0YkNOQT09


Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210316/5b8d23b5/attachment.html>

From shaojieb at andrew.cmu.edu  Wed Mar 17 09:40:04 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Wed, 17 Mar 2021 09:40:04 -0400
Subject: [UPDATE][CMU AI Seminar] Mar 24 (Zoom) -- Le Song (Georgia
 Tech/MBZUAI) -- Understanding Deep Architectures with Reasoning Layer -- AI
 Seminar sponsored by Fortive
Message-ID: <CAJA0PPCV2K1=kRto6Q3YCaY-U-uRT5D4Ce__5Hw5k2sorjDzhA@mail.gmail.com>

*UPDATE*: As the planned time (3/23 12pm) overlaps with a SCS faculty
candidate talk, this seminar will be moved to next Wednesday (3/24) at 12pm
instead.
--------------------------------------------

Dear all,

We look forward to seeing you *next Wednesday (3/24)* from 12:00-1:00 PM
(U.S. Eastern time) for the next talk of our *CMU AI seminar*, sponsored by
Fortive <https://careers.fortive.com/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
<http://www.cs.cmu.edu/~aiseminar/>

On 3/24, *Le Song* (Georgia Tech / MBZUAI) will be giving a talk on
"*Understanding
Deep Architectures with Reasoning Layer*."

*Title*: Understanding Deep Architectures with Reasoning Layer

*Talk Abstract*: Recently, there has been a surge of interest in combining
deep learning models with reasoning in order to handle more sophisticated
learning tasks. In many cases, a reasoning task can be solved by an
iterative algorithm. This algorithm is often unrolled, and used as a
specialized layer in the deep architecture, which can be trained end-to-end
with other neural components. Although such hybrid deep architectures have
led to many empirical successes, the theoretical foundation of such
architectures, especially the interplay between algorithm layers and other
neural layers, remains largely unexplored. In this paper, we take an
initial step towards an understanding of such hybrid deep architectures by
showing that properties of the algorithm layers, such as convergence,
stability and sensitivity, are intimately related to the approximation and
generalization abilities of the end-to-end model. Furthermore, our analysis
matches closely our experimental observations under various conditions,
suggesting that our theory can provide useful guidelines for designing deep
architectures with reasoning layers.

*Speaker Bio*: Le Song is a Professor and the Deputy Chair of the Machine
Learning Department, Mohamed bin Sayed University of AI, UAE. He was an
Associate Professor in the Department of Computational Science and
Engineering, College of Computing, Georgia Institute of Technology, worked
as a research scientist at Google, and did his post-doc in Carnegie Mellon
University. His principal research area is machine learning, especially
kernel methods, deep learning, and probabilistic graphical models. He is
the recipient of many best paper awards at major machine learning
conferences, such as NeurIPS, ICML and AISTATS, and the NSF CAREER Award.
He has also served as the area chair or senior program committee for many
leading machine learning and AI conferences such as NeurIPS, ICML, ICLR,
AAAI and IJCAI, and the action editor for JMLR and IEEE TPAMI. He is also a
board member of the International Conference for Machine Learning.

*Zoom Link*:
https://cmu.zoom.us/j/94092736800?pwd=OWlCRlV6RERtRHByWWpkZEl0YkNOQT09


Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210317/557b691f/attachment.html>

From shaojieb at andrew.cmu.edu  Tue Mar 23 12:06:47 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Tue, 23 Mar 2021 12:06:47 -0400
Subject: [UPDATE][CMU AI Seminar] Mar 24 (Zoom) -- Le Song (Georgia
 Tech/MBZUAI) -- Understanding Deep Architectures with Reasoning Layer -- AI
 Seminar sponsored by Fortive
In-Reply-To: <CAJA0PPCV2K1=kRto6Q3YCaY-U-uRT5D4Ce__5Hw5k2sorjDzhA@mail.gmail.com>
References: <CAJA0PPCV2K1=kRto6Q3YCaY-U-uRT5D4Ce__5Hw5k2sorjDzhA@mail.gmail.com>
Message-ID: <CAJA0PPBDfaGo-71uEyM4VOTc2qmDKJSqTaZk3wnTfKm+KeObWQ@mail.gmail.com>

Hi all,

*NOTE*: This seminar is tomorrow at *1pm*, not the usual 12pm (due to a
faculty candidate talk at 12pm).

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow *1pm-2pm*:
https://cmu.zoom.us/j/93418102649?pwd=TTd4dElxWnBOZHJ5QndUNVBWUjZCZz09.

Le Song (GeorgiaTech/MBZUAI) will be talking about some cool insights
related to deep reasoning layers (see below).

Thanks,
Shaojie


On Wed, Mar 17, 2021 at 9:40 AM Shaojie Bai <shaojieb at andrew.cmu.edu> wrote:

> *UPDATE*: As the planned time (3/23 12pm) overlaps with a SCS faculty
> candidate talk, this seminar will be moved to next Wednesday (3/24) at 12pm
> instead.
> --------------------------------------------
>
> Dear all,
>
> We look forward to seeing you *next Wednesday (3/24)* from 12:00-1:00 PM
> (U.S. Eastern time) for the next talk of our *CMU AI seminar*, sponsored
> by Fortive <https://careers.fortive.com/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
> <http://www.cs.cmu.edu/~aiseminar/>
>
> On 3/24, *Le Song* (Georgia Tech / MBZUAI) will be giving a talk on "*Understanding
> Deep Architectures with Reasoning Layer*."
>
> *Title*: Understanding Deep Architectures with Reasoning Layer
>
> *Talk Abstract*: Recently, there has been a surge of interest in
> combining deep learning models with reasoning in order to handle more
> sophisticated learning tasks. In many cases, a reasoning task can be solved
> by an iterative algorithm. This algorithm is often unrolled, and used as a
> specialized layer in the deep architecture, which can be trained end-to-end
> with other neural components. Although such hybrid deep architectures have
> led to many empirical successes, the theoretical foundation of such
> architectures, especially the interplay between algorithm layers and other
> neural layers, remains largely unexplored. In this paper, we take an
> initial step towards an understanding of such hybrid deep architectures by
> showing that properties of the algorithm layers, such as convergence,
> stability and sensitivity, are intimately related to the approximation and
> generalization abilities of the end-to-end model. Furthermore, our analysis
> matches closely our experimental observations under various conditions,
> suggesting that our theory can provide useful guidelines for designing deep
> architectures with reasoning layers.
>
> *Speaker Bio*: Le Song is a Professor and the Deputy Chair of the Machine
> Learning Department, Mohamed bin Sayed University of AI, UAE. He was an
> Associate Professor in the Department of Computational Science and
> Engineering, College of Computing, Georgia Institute of Technology, worked
> as a research scientist at Google, and did his post-doc in Carnegie Mellon
> University. His principal research area is machine learning, especially
> kernel methods, deep learning, and probabilistic graphical models. He is
> the recipient of many best paper awards at major machine learning
> conferences, such as NeurIPS, ICML and AISTATS, and the NSF CAREER Award.
> He has also served as the area chair or senior program committee for many
> leading machine learning and AI conferences such as NeurIPS, ICML, ICLR,
> AAAI and IJCAI, and the action editor for JMLR and IEEE TPAMI. He is also a
> board member of the International Conference for Machine Learning.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/94092736800?pwd=OWlCRlV6RERtRHByWWpkZEl0YkNOQT09
>
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210323/4f2440d2/attachment.html>

From shaojieb at andrew.cmu.edu  Tue Mar 23 13:16:29 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Tue, 23 Mar 2021 13:16:29 -0400
Subject: [UPDATE][CMU AI Seminar] Mar 24 (Zoom) -- Le Song (Georgia
 Tech/MBZUAI) -- Understanding Deep Architectures with Reasoning Layer -- AI
 Seminar sponsored by Fortive
In-Reply-To: <CAJA0PPBDfaGo-71uEyM4VOTc2qmDKJSqTaZk3wnTfKm+KeObWQ@mail.gmail.com>
References: <CAJA0PPCV2K1=kRto6Q3YCaY-U-uRT5D4Ce__5Hw5k2sorjDzhA@mail.gmail.com>
 <CAJA0PPBDfaGo-71uEyM4VOTc2qmDKJSqTaZk3wnTfKm+KeObWQ@mail.gmail.com>
Message-ID: <CAJA0PPA8Yd_Yq=ZWLpomXLv6zC2t7+-4Bsf-b8ghj-1PsqzdPw@mail.gmail.com>

Sorry for spamming everyone, it was pointed out to me that the Zoom link in
the last email is incorrect.

Here is the correct link:
https://cmu.zoom.us/j/94092736800?pwd=OWlCRlV6RERtRHByWWpkZEl0YkNOQT09.

*See you tomorrow (Wednesday) at 1pm ET for the talk with Le Song!*

On Tue, Mar 23, 2021 at 12:06 PM Shaojie Bai <shaojieb at andrew.cmu.edu>
wrote:

> Hi all,
>
> *NOTE*: This seminar is tomorrow at *1pm*, not the usual 12pm (due to a
> faculty candidate talk at 12pm).
>
> Just a reminder that the CMU AI Seminar
> <http://www.cs.cmu.edu/~aiseminar/> is tomorrow *1pm-2pm*:
> https://cmu.zoom.us/j/93418102649?pwd=TTd4dElxWnBOZHJ5QndUNVBWUjZCZz09.
>
> Le Song (GeorgiaTech/MBZUAI) will be talking about some cool insights
> related to deep reasoning layers (see below).
>
> Thanks,
> Shaojie
>
>
> On Wed, Mar 17, 2021 at 9:40 AM Shaojie Bai <shaojieb at andrew.cmu.edu>
> wrote:
>
>> *UPDATE*: As the planned time (3/23 12pm) overlaps with a SCS faculty
>> candidate talk, this seminar will be moved to next Wednesday (3/24) at 12pm
>> instead.
>> --------------------------------------------
>>
>> Dear all,
>>
>> We look forward to seeing you *next Wednesday (3/24)* from 12:00-1:00 PM
>> (U.S. Eastern time) for the next talk of our *CMU AI seminar*, sponsored
>> by Fortive <https://careers.fortive.com/>.
>>
>> To learn more about the seminar series or see the future schedule, please
>> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>> <http://www.cs.cmu.edu/~aiseminar/>
>>
>> On 3/24, *Le Song* (Georgia Tech / MBZUAI) will be giving a talk on "*Understanding
>> Deep Architectures with Reasoning Layer*."
>>
>> *Title*: Understanding Deep Architectures with Reasoning Layer
>>
>> *Talk Abstract*: Recently, there has been a surge of interest in
>> combining deep learning models with reasoning in order to handle more
>> sophisticated learning tasks. In many cases, a reasoning task can be solved
>> by an iterative algorithm. This algorithm is often unrolled, and used as a
>> specialized layer in the deep architecture, which can be trained end-to-end
>> with other neural components. Although such hybrid deep architectures have
>> led to many empirical successes, the theoretical foundation of such
>> architectures, especially the interplay between algorithm layers and other
>> neural layers, remains largely unexplored. In this paper, we take an
>> initial step towards an understanding of such hybrid deep architectures by
>> showing that properties of the algorithm layers, such as convergence,
>> stability and sensitivity, are intimately related to the approximation and
>> generalization abilities of the end-to-end model. Furthermore, our analysis
>> matches closely our experimental observations under various conditions,
>> suggesting that our theory can provide useful guidelines for designing deep
>> architectures with reasoning layers.
>>
>> *Speaker Bio*: Le Song is a Professor and the Deputy Chair of the
>> Machine Learning Department, Mohamed bin Sayed University of AI, UAE. He
>> was an Associate Professor in the Department of Computational Science and
>> Engineering, College of Computing, Georgia Institute of Technology, worked
>> as a research scientist at Google, and did his post-doc in Carnegie Mellon
>> University. His principal research area is machine learning, especially
>> kernel methods, deep learning, and probabilistic graphical models. He is
>> the recipient of many best paper awards at major machine learning
>> conferences, such as NeurIPS, ICML and AISTATS, and the NSF CAREER Award.
>> He has also served as the area chair or senior program committee for many
>> leading machine learning and AI conferences such as NeurIPS, ICML, ICLR,
>> AAAI and IJCAI, and the action editor for JMLR and IEEE TPAMI. He is also a
>> board member of the International Conference for Machine Learning.
>>
>> *Zoom Link*:
>> https://cmu.zoom.us/j/94092736800?pwd=OWlCRlV6RERtRHByWWpkZEl0YkNOQT09
>>
>>
>> Thanks,
>> Shaojie Bai (MLD)
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210323/b383fd43/attachment-0001.html>

From shaojieb at andrew.cmu.edu  Wed Mar 24 11:51:02 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Wed, 24 Mar 2021 11:51:02 -0400
Subject: [CMU AI Seminar] - TODAY at 1pm (Zoom) -- Le Song (Georgia
 Tech/MBZUAI) -- Understanding Deep Architectures with Reasoning Layer -- AI
 Seminar sponsored by Fortive
Message-ID: <CAJA0PPAt-CfX=4_rXvGZ2FfHfw4ohhrX=pJwRnDCx1DttPT_dw@mail.gmail.com>

Dear all,

Just want to remind everyone again that this week's CMU AI seminar is *today
(3/24)* *from 1pm-2pm*.

*Le Song* (Georgia Tech / MBZUAI) will be giving a talk on "*Understanding
Deep Architectures with Reasoning Layer*."

*Title*: Understanding Deep Architectures with Reasoning Layer

*Talk Abstract*: Recently, there has been a surge of interest in combining
deep learning models with the reasoning in order to handle more
sophisticated learning tasks. In many cases, a reasoning task can be solved
by an iterative algorithm. This algorithm is often unrolled, and used as a
specialized layer in the deep architecture, which can be trained end-to-end
with other neural components. Although such hybrid deep architectures have
led to many empirical successes, the theoretical foundation of such
architectures, especially the interplay between algorithm layers and other
neural layers, remains largely unexplored. In this paper, we take an
initial step towards an understanding of such hybrid deep architectures by
showing that properties of the algorithm layers, such as convergence,
stability and sensitivity, are intimately related to the approximation and
generalization abilities of the end-to-end model. Furthermore, our analysis
matches closely our experimental observations under various conditions,
suggesting that our theory can provide useful guidelines for designing deep
architectures with reasoning layers.

*Speaker Bio*: Le Song is a Professor and the Deputy Chair of the Machine
Learning Department, Mohamed bin Sayed University of AI, UAE. He was an
Associate Professor in the Department of Computational Science and
Engineering, College of Computing, Georgia Institute of Technology, worked
as a research scientist at Google, and did his post-doc in Carnegie Mellon
University. His principal research area is machine learning, especially
kernel methods, deep learning, and probabilistic graphical models. He is
the recipient of many best paper awards at major machine learning
conferences, such as NeurIPS, ICML and AISTATS, and the NSF CAREER Award.
He has also served as the area chair or senior program committee for many
leading machine learning and AI conferences such as NeurIPS, ICML, ICLR,
AAAI and IJCAI, and the action editor for JMLR and IEEE TPAMI. He is also a
board member of the International Conference for Machine Learning.

*Zoom Link*:
https://cmu.zoom.us/j/94092736800?pwd=OWlCRlV6RERtRHByWWpkZEl0YkNOQT09


Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210324/a64d6254/attachment.html>

From shaojieb at andrew.cmu.edu  Wed Mar 24 14:48:00 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Wed, 24 Mar 2021 14:48:00 -0400
Subject: [CMU AI Seminar] Mar 30 at 2pm (Zoom) -- Been Kim (Google Brain) --
 AI Interpretability: the Past,
 Present and Future -- AI Seminar sponsored by Fortive
Message-ID: <CAJA0PPB8FhZM2Z04bbyNAE8RLqxKcskdyeKw5LoLZbSTVP88Tg@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (3/30)* from *2:00-3:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
by Fortive <https://careers.fortive.com/>. (Note: not the usual 12pm!)

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
<http://www.cs.cmu.edu/~aiseminar/>

On 3/30, *Been Kim* (Google Brain) will be giving a talk on "*AI
Interpretability: the Past, Present and Future*."

*Title*: AI Interpretability: the Past, Present and Future

*Talk Abstract*: Interpretable machine learning has been a popular topic of
study in the past many years. But are we making progress? In this talk, I
will talk about my reflections on the progress by taking a critical look at
some of the existing methods, and discussing series of user-centric methods
that can "speak" the user's language, rather than the computer's language.

*Speaker Bio*: Been Kim is a staff research scientist at Google Brain. Her
research focuses on improving interpretability in machine learning by
building interpretability methods for already-trained models or building
inherently interpretable models. She gave a talk at the G20 meeting in
Argentina in 2019. Her work TCAV received UNESCO Netexplo award, was
featured at Google I/O 19' and in Brian Christian's book on "The Alignment
Problem". Been has given keynote at ECML 2020, tutorials on
interpretability at ICML, University of Toronto, CVPR and at Lawrence
Berkeley National Laboratory. She was a co-workshop Chair ICLR 2019, and
has been an area chair/senior area chair at conferences including NeurIPS,
ICML, ICLR, and AISTATS. She received her Ph.D. from MIT.

*Zoom Link*:
https://cmu.zoom.us/j/92759487765?pwd=Zzl4Ui9LU0cwMFdRTkYyZDdZQ2MyQT09


Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210324/3b8cb215/attachment.html>

From shaojieb at andrew.cmu.edu  Mon Mar 29 13:10:47 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Mon, 29 Mar 2021 13:10:47 -0400
Subject: [CMU AI Seminar] Mar 30 at 2pm (Zoom) -- Been Kim (Google Brain)
 -- AI Interpretability: the Past,
 Present and Future -- AI Seminar sponsored by Fortive
In-Reply-To: <CAJA0PPB8FhZM2Z04bbyNAE8RLqxKcskdyeKw5LoLZbSTVP88Tg@mail.gmail.com>
References: <CAJA0PPB8FhZM2Z04bbyNAE8RLqxKcskdyeKw5LoLZbSTVP88Tg@mail.gmail.com>
Message-ID: <CAJA0PPB+OqYd6jrWO4RRfOOxcm3cN20YPg6UppmH1fK-t0fBNA@mail.gmail.com>

Dear all,

Hi all,

*NOTE*: This seminar is tomorrow at *2pm*, not the usual 12pm (due to a
faculty job talk at 12pm).

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow *2pm-3pm*:
https://cmu.zoom.us/j/92759487765?pwd=Zzl4Ui9LU0cwMFdRTkYyZDdZQ2MyQT09.

Been Kim (Google Brain) will be talking about her visions on Explainable AI.

Thanks,
Shaojie

On Wed, Mar 24, 2021 at 2:48 PM Shaojie Bai <shaojieb at andrew.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (3/30)* from *2:00-3:00 PM
> (U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
> by Fortive <https://careers.fortive.com/>. (Note: not the usual 12pm!)
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
> <http://www.cs.cmu.edu/~aiseminar/>
>
> On 3/30, *Been Kim* (Google Brain) will be giving a talk on "*AI
> Interpretability: the Past, Present and Future*."
>
> *Title*: AI Interpretability: the Past, Present and Future
>
> *Talk Abstract*: Interpretable machine learning has been a popular topic
> of study in the past many years. But are we making progress? In this talk,
> I will talk about my reflections on the progress by taking a critical look
> at some of the existing methods, and discussing series of user-centric
> methods that can "speak" the user's language, rather than the computer's
> language.
>
> *Speaker Bio*: Been Kim is a staff research scientist at Google Brain.
> Her research focuses on improving interpretability in machine learning by
> building interpretability methods for already-trained models or building
> inherently interpretable models. She gave a talk at the G20 meeting in
> Argentina in 2019. Her work TCAV received UNESCO Netexplo award, was
> featured at Google I/O 19' and in Brian Christian's book on "The Alignment
> Problem". Been has given keynote at ECML 2020, tutorials on
> interpretability at ICML, University of Toronto, CVPR and at Lawrence
> Berkeley National Laboratory. She was a co-workshop Chair ICLR 2019, and
> has been an area chair/senior area chair at conferences including NeurIPS,
> ICML, ICLR, and AISTATS. She received her Ph.D. from MIT.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/92759487765?pwd=Zzl4Ui9LU0cwMFdRTkYyZDdZQ2MyQT09
>
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210329/5efd75bc/attachment.html>

From shaojieb at andrew.cmu.edu  Tue Mar 30 16:14:37 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Tue, 30 Mar 2021 16:14:37 -0400
Subject: [CMU AI Seminar] Apr 6 at 12pm (Zoom) -- Elan Rosenfeld (CMU MLD) --
 The Risks of Invariant Risk Minimization -- AI Seminar sponsored by
 Fortive
Message-ID: <CAJA0PPAnajrLjft_r8Q+RzynTM0M2cA3wtoYQ1Xnq3voez_m1Q@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (4/6)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
by Fortive <https://careers.fortive.com/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
<http://www.cs.cmu.edu/~aiseminar/>

On 4/6, *Elan Rosenfeld* (CMU MLD) will be giving a talk on "*The Risks of
Invariant Risk Minimization*."

*Title*: The Risks of Invariant Risk Minimization

*Talk Abstract*: Invariant feature learning has become a popular
alternative to Empirical Risk Minimization as practitioners recognize the
need to ignore features which may be misleading at test time in order to
improve out-of-distribution generalization. Early results in this area
leverage variation across environments to provably identify the features
which are directly causal with respect to the target variable. More recent
work attempts to use this technique for deep learning, frequently with no
formal guarantees of an algorithm's ability to uncover the correct
features. Most notably, the seminal work introducing Invariant Risk
Minimization gave a loose bound for the linear setting and no results at
all for non-linearity; despite this, a large number of variations have been
suggested. In this talk, I'll introduce a formal latent variable model
which encodes the primary assumptions made by these works. I'll then give
the first characterization of the optimal solution to the IRM objective,
deriving the exact number of environments needed for the solution to
generalize in the linear case. Finally, I'll present the first analysis of
IRM when the observed data is a non-linear function of the latent
variables: in particular, we show that IRM can fail catastrophically when
the test distribution is even moderately different from the training
distribution - this is exactly the problem that IRM was intended to solve.
These results easily generalize to all recent variations on IRM,
demonstrating that these works on invariant feature learning fundamentally
do not improve over standard ERM. This talk is based on work with Pradeep
Ravikumar and Andrej Risteski, to appear at ICLR 2021.

*Speaker Bio*: Elan Rosenfeld is a PhD student in the Machine Learning
Department at CMU, advised by Andrej Risteski and Pradeep Ravikumar. He is
interested in theoretical foundations of machine learning, with a
particular focus on robust learning, representation learning and
out-of-distribution generalization. Elan completed his undergraduate
degrees in Computer Science and Statistics & Machine Learning at CMU, where
his senior thesis on human-usable password schemas was advised by Manuel
Blum and Santosh Vempala.

*Zoom Link*:
https://cmu.zoom.us/j/96139997861?pwd=ZlMrUUZaWXY0Sm9mai9ZdjE4QXNDQT09


Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210330/3523507e/attachment.html>

From shaojieb at andrew.cmu.edu  Mon Apr  5 13:26:59 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Mon, 5 Apr 2021 13:26:59 -0400
Subject: [CMU AI Seminar] Apr 6 at 12pm (Zoom) -- Elan Rosenfeld (CMU MLD)
 -- The Risks of Invariant Risk Minimization -- AI Seminar sponsored
 by Fortive
In-Reply-To: <CAJA0PPAnajrLjft_r8Q+RzynTM0M2cA3wtoYQ1Xnq3voez_m1Q@mail.gmail.com>
References: <CAJA0PPAnajrLjft_r8Q+RzynTM0M2cA3wtoYQ1Xnq3voez_m1Q@mail.gmail.com>
Message-ID: <CAJA0PPB98kQpkOk-Y28vZ8_+2VeKHQ0VqP5BqvJVmOWK0En6gg@mail.gmail.com>

Dear all,

*NOTE/UPDATE*: This seminar is tomorrow at *1pm* (due to a faculty job talk
at 12pm).

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow *1pm-2pm*:
https://cmu.zoom.us/j/96139997861?pwd=ZlMrUUZaWXY0Sm9mai9ZdjE4QXNDQT09.

Elan Rosenfeld (CMU MLD) will be talking about his latest work on invariant
risk minimization (IRM).

Thanks,
Shaojie

On Tue, Mar 30, 2021 at 4:14 PM Shaojie Bai <shaojieb at andrew.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (4/6)* from *1**2:00-1:00 PM
> (U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
> by Fortive <https://careers.fortive.com/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
> <http://www.cs.cmu.edu/~aiseminar/>
>
> On 4/6, *Elan Rosenfeld* (CMU MLD) will be giving a talk on "*The Risks
> of Invariant Risk Minimization*."
>
> *Title*: The Risks of Invariant Risk Minimization
>
> *Talk Abstract*: Invariant feature learning has become a popular
> alternative to Empirical Risk Minimization as practitioners recognize the
> need to ignore features which may be misleading at test time in order to
> improve out-of-distribution generalization. Early results in this area
> leverage variation across environments to provably identify the features
> which are directly causal with respect to the target variable. More recent
> work attempts to use this technique for deep learning, frequently with no
> formal guarantees of an algorithm's ability to uncover the correct
> features. Most notably, the seminal work introducing Invariant Risk
> Minimization gave a loose bound for the linear setting and no results at
> all for non-linearity; despite this, a large number of variations have been
> suggested. In this talk, I'll introduce a formal latent variable model
> which encodes the primary assumptions made by these works. I'll then give
> the first characterization of the optimal solution to the IRM objective,
> deriving the exact number of environments needed for the solution to
> generalize in the linear case. Finally, I'll present the first analysis of
> IRM when the observed data is a non-linear function of the latent
> variables: in particular, we show that IRM can fail catastrophically when
> the test distribution is even moderately different from the training
> distribution - this is exactly the problem that IRM was intended to solve.
> These results easily generalize to all recent variations on IRM,
> demonstrating that these works on invariant feature learning fundamentally
> do not improve over standard ERM. This talk is based on work with Pradeep
> Ravikumar and Andrej Risteski, to appear at ICLR 2021.
>
> *Speaker Bio*: Elan Rosenfeld is a PhD student in the Machine Learning
> Department at CMU, advised by Andrej Risteski and Pradeep Ravikumar. He is
> interested in theoretical foundations of machine learning, with a
> particular focus on robust learning, representation learning and
> out-of-distribution generalization. Elan completed his undergraduate
> degrees in Computer Science and Statistics & Machine Learning at CMU, where
> his senior thesis on human-usable password schemas was advised by Manuel
> Blum and Santosh Vempala.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/96139997861?pwd=ZlMrUUZaWXY0Sm9mai9ZdjE4QXNDQT09
>
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210405/945f41a6/attachment.html>

From shaojieb at andrew.cmu.edu  Tue Apr  6 13:08:12 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Tue, 6 Apr 2021 13:08:12 -0400
Subject: [CMU AI Seminar] Apr 6 at 12pm (Zoom) -- Elan Rosenfeld (CMU MLD)
 -- The Risks of Invariant Risk Minimization -- AI Seminar sponsored
 by Fortive
In-Reply-To: <CAJA0PPB98kQpkOk-Y28vZ8_+2VeKHQ0VqP5BqvJVmOWK0En6gg@mail.gmail.com>
References: <CAJA0PPAnajrLjft_r8Q+RzynTM0M2cA3wtoYQ1Xnq3voez_m1Q@mail.gmail.com>
 <CAJA0PPB98kQpkOk-Y28vZ8_+2VeKHQ0VqP5BqvJVmOWK0En6gg@mail.gmail.com>
Message-ID: <CAJA0PPDTf4vLnr5VM2b2OONa+YCQzEKc7s0hCrykVOU_b2W2BA@mail.gmail.com>

Reminder: This talk is happening now! Please join us at
https://cmu.zoom.us/j/96139997861?pwd=ZlMrUUZaWXY0Sm9mai9ZdjE4QXNDQT09.

On Mon, Apr 5, 2021 at 1:26 PM Shaojie Bai <shaojieb at andrew.cmu.edu> wrote:

> Dear all,
>
> *NOTE/UPDATE*: This seminar is tomorrow at *1pm* (due to a faculty job
> talk at 12pm).
>
> Just a reminder that the CMU AI Seminar
> <http://www.cs.cmu.edu/~aiseminar/> is tomorrow *1pm-2pm*:
> https://cmu.zoom.us/j/96139997861?pwd=ZlMrUUZaWXY0Sm9mai9ZdjE4QXNDQT09.
>
> Elan Rosenfeld (CMU MLD) will be talking about his latest work on
> invariant risk minimization (IRM).
>
> Thanks,
> Shaojie
>
> On Tue, Mar 30, 2021 at 4:14 PM Shaojie Bai <shaojieb at andrew.cmu.edu>
> wrote:
>
>> Dear all,
>>
>> We look forward to seeing you *next Tuesday (4/6)* from *1**2:00-1:00 PM
>> (U.S. Eastern time)* for the next talk of our *CMU AI seminar*,
>> sponsored by Fortive <https://careers.fortive.com/>.
>>
>> To learn more about the seminar series or see the future schedule, please
>> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>> <http://www.cs.cmu.edu/~aiseminar/>
>>
>> On 4/6, *Elan Rosenfeld* (CMU MLD) will be giving a talk on "*The Risks
>> of Invariant Risk Minimization*."
>>
>> *Title*: The Risks of Invariant Risk Minimization
>>
>> *Talk Abstract*: Invariant feature learning has become a popular
>> alternative to Empirical Risk Minimization as practitioners recognize the
>> need to ignore features which may be misleading at test time in order to
>> improve out-of-distribution generalization. Early results in this area
>> leverage variation across environments to provably identify the features
>> which are directly causal with respect to the target variable. More recent
>> work attempts to use this technique for deep learning, frequently with no
>> formal guarantees of an algorithm's ability to uncover the correct
>> features. Most notably, the seminal work introducing Invariant Risk
>> Minimization gave a loose bound for the linear setting and no results at
>> all for non-linearity; despite this, a large number of variations have been
>> suggested. In this talk, I'll introduce a formal latent variable model
>> which encodes the primary assumptions made by these works. I'll then give
>> the first characterization of the optimal solution to the IRM objective,
>> deriving the exact number of environments needed for the solution to
>> generalize in the linear case. Finally, I'll present the first analysis of
>> IRM when the observed data is a non-linear function of the latent
>> variables: in particular, we show that IRM can fail catastrophically when
>> the test distribution is even moderately different from the training
>> distribution - this is exactly the problem that IRM was intended to solve.
>> These results easily generalize to all recent variations on IRM,
>> demonstrating that these works on invariant feature learning fundamentally
>> do not improve over standard ERM. This talk is based on work with Pradeep
>> Ravikumar and Andrej Risteski, to appear at ICLR 2021.
>>
>> *Speaker Bio*: Elan Rosenfeld is a PhD student in the Machine Learning
>> Department at CMU, advised by Andrej Risteski and Pradeep Ravikumar. He is
>> interested in theoretical foundations of machine learning, with a
>> particular focus on robust learning, representation learning and
>> out-of-distribution generalization. Elan completed his undergraduate
>> degrees in Computer Science and Statistics & Machine Learning at CMU, where
>> his senior thesis on human-usable password schemas was advised by Manuel
>> Blum and Santosh Vempala.
>>
>> *Zoom Link*:
>> https://cmu.zoom.us/j/96139997861?pwd=ZlMrUUZaWXY0Sm9mai9ZdjE4QXNDQT09
>>
>>
>> Thanks,
>> Shaojie Bai (MLD)
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210406/f8cf7a10/attachment.html>

From shaojieb at andrew.cmu.edu  Tue Apr  6 14:10:18 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Tue, 6 Apr 2021 14:10:18 -0400
Subject: [CMU AI Seminar] Apr 13 at 12pm (Zoom) -- Noah Smith (U of
 Washington) -- Language Models: Challenges and Progress -- AI Seminar
 sponsored by Fortive
Message-ID: <CAJA0PPBmk8gpF6cykF0D+YU9SJ9A9WDWq0xKTgbYn4pOeqMQGg@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (4/13)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
by Fortive <https://careers.fortive.com/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
<http://www.cs.cmu.edu/~aiseminar/>

On 4/13, *Noah Smith* (University of Washington / AI2) will be giving a
talk on "*Language Models: Challenges and Progress*".

*Title*: Language Models: Challenges and Progress

*Talk Abstract*: Probabilistic language models are once again foundational
to many advances in natural language processing research, bringing the
exciting opportunity to harness raw text to build language technologies.
With the emergence of deep architectures and protocols for finetuning a
pretrained language model, many NLP solutions are being cast as simple
variations on language modeling. This talk is about challenges in language
model-based NLP and some of our work toward solutions. First, we'll
consider evaluation of generated language. I'll present some alarming
findings about humans and models and make some recommendations. Second,
I'll turn to an ubiquitous design limitation in language modeling -- the
vocabulary -- and present a linguistically principled, sample-efficient
solution that enables modifying the vocabulary during finetuning and/or
deployment. Finally, I'll delve into today's most popular language modeling
architecture, the transformer, and show how its attention layers' quadratic
runtime can be made linear without affecting accuracy. Taken together, we
hope these advances will broaden the population of people who can
effectively use and contribute back to NLP.

*Speaker Bio*: Noah Smith is a Professor in the Paul G. Allen School of
Computer Science & Engineering at the University of Washington, as well as
a Senior Research Manager at the Allen Institute for Artificial
Intelligence. Previously, he was an Associate Professor of Language
Technologies and Machine Learning in the School of Computer Science at
Carnegie Mellon University. He received his Ph.D. in Computer Science from
Johns Hopkins University in 2006 and his B.S. in Computer Science and B.A.
in Linguistics from the University of Maryland in 2001. His research
interests include statistical natural language processing, machine
learning, and applications of natural language processing, especially to
the social sciences. His book, Linguistic Structure Prediction, covers many
of these topics. He has served on the editorial boards of the journals
Computational Linguistics (2009-2011), Journal of Artificial Intelligence
Research (2011-present), and Transactions of the Association for
Computational Linguistics (2012-present), as the secretary-treasurer of
SIGDAT (2012-2015 and 2018-present), and as program co-chair of ACL 2016.
Alumni of his research group, Noah's ARK, are international leaders in NLP
in academia and industry; in 2017 UW's Sounding Board team won the
inaugural Amazon Alexa Prize. He was named an ACL Fellow in 2020, "for
significant contributions to linguistic structure prediction, computational
social sciences, and improving NLP research methodology." Smith's work has
been recognized with a UW Innovation award (2016-2018), a Finmeccanica
career development chair at CMU (2011-2014), an NSF CAREER award
(2011-2016), a Hertz Foundation graduate fellowship (2001-2006), numerous
best paper nominations and awards, and coverage by NPR, BBC, CBC, New York
Times, Washington Post, and Time.

*Zoom Link*:
https://cmu.zoom.us/j/93338025712?pwd=dEZvTkc0bTVtTjNkRkQzeGo5KzVZUT09

Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210406/29563cca/attachment.html>

From shaojieb at andrew.cmu.edu  Mon Apr 12 12:28:01 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Mon, 12 Apr 2021 12:28:01 -0400
Subject: [CMU AI Seminar] Apr 13 at 12pm (Zoom) -- Noah Smith (U of
 Washington) -- Language Models: Challenges and Progress -- AI Seminar
 sponsored by Fortive
In-Reply-To: <CAJA0PPBmk8gpF6cykF0D+YU9SJ9A9WDWq0xKTgbYn4pOeqMQGg@mail.gmail.com>
References: <CAJA0PPBmk8gpF6cykF0D+YU9SJ9A9WDWq0xKTgbYn4pOeqMQGg@mail.gmail.com>
Message-ID: <CAJA0PPCO10ZCJtz+JXy0W+MVh4ogZAObwRqw=K5yfWP2ALY9WA@mail.gmail.com>

Hi all,

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow *12pm-1pm*:
https://cmu.zoom.us/j/93338025712?pwd=dEZvTkc0bTVtTjNkRkQzeGo5KzVZUT09.

*Noah Smith (University of Washington / AI2)* will be talking about some
latest progress in language modeling!

Thanks,
Shaojie

On Tue, Apr 6, 2021 at 2:10 PM Shaojie Bai <shaojieb at andrew.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (4/13)* from *1**2:00-1:00 PM
> (U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
> by Fortive <https://careers.fortive.com/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
> <http://www.cs.cmu.edu/~aiseminar/>
>
> On 4/13, *Noah Smith* (University of Washington / AI2) will be giving a
> talk on "*Language Models: Challenges and Progress*".
>
> *Title*: Language Models: Challenges and Progress
>
> *Talk Abstract*: Probabilistic language models are once again
> foundational to many advances in natural language processing research,
> bringing the exciting opportunity to harness raw text to build language
> technologies. With the emergence of deep architectures and protocols for
> finetuning a pretrained language model, many NLP solutions are being cast
> as simple variations on language modeling. This talk is about challenges in
> language model-based NLP and some of our work toward solutions. First,
> we'll consider evaluation of generated language. I'll present some alarming
> findings about humans and models and make some recommendations. Second,
> I'll turn to an ubiquitous design limitation in language modeling -- the
> vocabulary -- and present a linguistically principled, sample-efficient
> solution that enables modifying the vocabulary during finetuning and/or
> deployment. Finally, I'll delve into today's most popular language modeling
> architecture, the transformer, and show how its attention layers' quadratic
> runtime can be made linear without affecting accuracy. Taken together, we
> hope these advances will broaden the population of people who can
> effectively use and contribute back to NLP.
>
> *Speaker Bio*: Noah Smith is a Professor in the Paul G. Allen School of
> Computer Science & Engineering at the University of Washington, as well as
> a Senior Research Manager at the Allen Institute for Artificial
> Intelligence. Previously, he was an Associate Professor of Language
> Technologies and Machine Learning in the School of Computer Science at
> Carnegie Mellon University. He received his Ph.D. in Computer Science from
> Johns Hopkins University in 2006 and his B.S. in Computer Science and B.A.
> in Linguistics from the University of Maryland in 2001. His research
> interests include statistical natural language processing, machine
> learning, and applications of natural language processing, especially to
> the social sciences. His book, Linguistic Structure Prediction, covers many
> of these topics. He has served on the editorial boards of the journals
> Computational Linguistics (2009-2011), Journal of Artificial Intelligence
> Research (2011-present), and Transactions of the Association for
> Computational Linguistics (2012-present), as the secretary-treasurer of
> SIGDAT (2012-2015 and 2018-present), and as program co-chair of ACL 2016.
> Alumni of his research group, Noah's ARK, are international leaders in NLP
> in academia and industry; in 2017 UW's Sounding Board team won the
> inaugural Amazon Alexa Prize. He was named an ACL Fellow in 2020, "for
> significant contributions to linguistic structure prediction, computational
> social sciences, and improving NLP research methodology." Smith's work has
> been recognized with a UW Innovation award (2016-2018), a Finmeccanica
> career development chair at CMU (2011-2014), an NSF CAREER award
> (2011-2016), a Hertz Foundation graduate fellowship (2001-2006), numerous
> best paper nominations and awards, and coverage by NPR, BBC, CBC, New York
> Times, Washington Post, and Time.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/93338025712?pwd=dEZvTkc0bTVtTjNkRkQzeGo5KzVZUT09
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210412/129e590a/attachment.html>

From shaojieb at andrew.cmu.edu  Thu Apr 15 12:22:09 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Thu, 15 Apr 2021 12:22:09 -0400
Subject: [CMU AI Seminar] Apr 20 at 12pm (Zoom) -- Misha Khodak (CMU) --
 Factorized Layers Revisited: Compressing Deep Neural Networks Without Playing
 the Lottery -- AI Seminar sponsored by Fortive
Message-ID: <CAJA0PPAW1+H51k-4JcXMubPs08YwwajEBy6HxswjrxSsQcK3RQ@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (4/20)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
by Fortive <https://careers.fortive.com/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
<http://www.cs.cmu.edu/~aiseminar/>

On 4/20, *Misha Khodak* (CMU CSD) will be giving a talk on "*Factorized
Layers Revisited: Compressing Deep Neural Networks Without Playing the
Lottery*".

*Title*: Factorized Layers Revisited: Compressing Deep Neural Networks
Without Playing the Lottery

*Talk Abstract*: Machine learning models are rapidly growing in size,
leading to increased training and deployment costs. While the most popular
approach for training compressed models is trying to guess good "lottery
tickets" or sparse subnetworks, we revisit the low-rank factorization
approach, in which weights matrices are replaced by products of smaller
matrices. We extend recent analyses of optimization of deep networks to
motivate simple initialization and regularization schemes for improving the
training of these factorized layers. Empirically these methods yield higher
accuracies than popular pruning and lottery ticket approaches at the same
compression level. We further demonstrate their usefulness in two settings
beyond model compression: simplifying knowledge distillation and training
Transformer-based architectures such as BERT. This is joint work with Neil
Tenenholtz, Lester Mackey, and Nicolo Fusi.

*Speaker Bio*: Misha Khodaka is a PhD student in Carnegie Mellon
University's Computer Science Department advised by Nina Balcan and Ameet
Talwalkar. His research focuses on foundations and applications of machine
learning, most recently neural architecture search, meta-learning, and
unsupervised representation learning. He recently spent time as an intern
with Nicolo Fusi at Microsoft Research - New England and previously
received an AB in Mathematics and an MSE in Computer Science from Princeton
University, where he worked with Sanjeev Arora.

*Zoom Link*:
https://cmu.zoom.us/j/93099996457?pwd=b3BSSHp2RWZWQjZ0SUE4ZkdKSDk4UT09

Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210415/2ee69d68/attachment.html>

From shaojieb at andrew.cmu.edu  Mon Apr 19 14:10:28 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Mon, 19 Apr 2021 14:10:28 -0400
Subject: [CMU AI Seminar] Apr 20 at 12pm (Zoom) -- Misha Khodak (CMU) --
 Factorized Layers Revisited: Compressing Deep Neural Networks Without Playing
 the Lottery -- AI Seminar sponsored by Fortive
In-Reply-To: <CAJA0PPAW1+H51k-4JcXMubPs08YwwajEBy6HxswjrxSsQcK3RQ@mail.gmail.com>
References: <CAJA0PPAW1+H51k-4JcXMubPs08YwwajEBy6HxswjrxSsQcK3RQ@mail.gmail.com>
Message-ID: <CAJA0PPD4=PXLk7p+z2faqBRVzw9fee-xVx9kAXUqLJdXrCQMrg@mail.gmail.com>

Dear all,

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow *1**2pm-1pm*:
https://cmu.zoom.us/j/93099996457?pwd=b3BSSHp2RWZWQjZ0SUE4ZkdKSDk4UT09.

Misha Khodak (CMU CSD) will be talking about his recent ICLR work on model
compression with layer factorization.

Thanks,
Shaojie

On Thu, Apr 15, 2021 at 12:22 PM Shaojie Bai <shaojieb at andrew.cmu.edu>
wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (4/20)* from *1**2:00-1:00 PM
> (U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
> by Fortive <https://careers.fortive.com/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
> <http://www.cs.cmu.edu/~aiseminar/>
>
> On 4/20, *Misha Khodak* (CMU CSD) will be giving a talk on "*Factorized
> Layers Revisited: Compressing Deep Neural Networks Without Playing the
> Lottery*".
>
> *Title*: Factorized Layers Revisited: Compressing Deep Neural Networks
> Without Playing the Lottery
>
> *Talk Abstract*: Machine learning models are rapidly growing in size,
> leading to increased training and deployment costs. While the most popular
> approach for training compressed models is trying to guess good "lottery
> tickets" or sparse subnetworks, we revisit the low-rank factorization
> approach, in which weights matrices are replaced by products of smaller
> matrices. We extend recent analyses of optimization of deep networks to
> motivate simple initialization and regularization schemes for improving the
> training of these factorized layers. Empirically these methods yield higher
> accuracies than popular pruning and lottery ticket approaches at the same
> compression level. We further demonstrate their usefulness in two settings
> beyond model compression: simplifying knowledge distillation and training
> Transformer-based architectures such as BERT. This is joint work with Neil
> Tenenholtz, Lester Mackey, and Nicolo Fusi.
>
> *Speaker Bio*: Misha Khodaka is a PhD student in Carnegie Mellon
> University's Computer Science Department advised by Nina Balcan and Ameet
> Talwalkar. His research focuses on foundations and applications of machine
> learning, most recently neural architecture search, meta-learning, and
> unsupervised representation learning. He recently spent time as an intern
> with Nicolo Fusi at Microsoft Research - New England and previously
> received an AB in Mathematics and an MSE in Computer Science from Princeton
> University, where he worked with Sanjeev Arora.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/93099996457?pwd=b3BSSHp2RWZWQjZ0SUE4ZkdKSDk4UT09
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210419/a61b5894/attachment.html>

From shaojieb at andrew.cmu.edu  Tue Apr 20 13:15:34 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Tue, 20 Apr 2021 13:15:34 -0400
Subject: [CMU AI Seminar] Apr 27 at 12pm (Zoom) -- Bo Li (UIUC) -- Secure
 Learning in Adversarial Environments -- AI Seminar sponsored by Fortive
Message-ID: <CAJA0PPDBYj-1cMXfzFk36k2FZKONbena9-bcOKtSGFVJsuHtbQ@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (4/27)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
by Fortive <https://careers.fortive.com/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
<http://www.cs.cmu.edu/~aiseminar/>

On 4/27, *Bo Li* (UIUC) will be giving a talk on "*Secure Learning in
Adversarial Environments*".

*Title*: Secure Learning in Adversarial Environments

*Talk Abstract*: Advances in machine learning have led to rapid and
widespread deployment of learning based inference and decision making for
safety-critical applications, such as autonomous driving and security
diagnostics. Current machine learning systems, however, assume that
training and test data follow the same, or similar, distributions, and do
not consider active adversaries manipulating either distribution. Recent
work has demonstrated that motivated adversaries can circumvent anomaly
detection or other machine learning models at test time through evasion
attacks, or can inject well-crafted malicious instances into training data
to induce errors in inference time through poisoning attacks. In this talk,
I will describe my recent research about security and privacy problems in
machine learning systems. In particular, I will introduce several
adversarial attacks in different domains, and discuss potential defensive
approaches and principles, including game theoretic based and knowledge
enabled robust learning paradigms, towards developing practical robust
learning systems with robustness guarantees.

*Speaker Bio*: Dr. Bo Li is an assistant professor in the department of
Computer Science at University of Illinois at Urbana-Champaign, and the
recipient of the Symantec Research Labs Fellowship, Rising Stars, MIT
Technology Review TR-35 award, Intel Rising Star award, Amazon Research
Award, and best paper awards in several machine learning and security
conferences. Previously she was a postdoctoral researcher in UC Berkeley.
Her research focuses on both theoretical and practical aspects of security,
machine learning, privacy, game theory, and adversarial machine learning.
She has designed several robust learning algorithms, scalable frameworks
for achieving robustness for a range of learning methods, and a privacy
preserving data publishing system. Her work have been featured by major
publications and media outlets such as Nature, Wired, Fortune, and New York
Times.

*Zoom Link*:
https://cmu.zoom.us/j/92752459790?pwd=aWN0NXhxaitPTHlnaEZpYUFuNUk1UT09

Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210420/9d0410c3/attachment.html>

From shaojieb at andrew.cmu.edu  Mon Apr 26 15:46:59 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Mon, 26 Apr 2021 15:46:59 -0400
Subject: [CMU AI Seminar] Apr 27 at 12pm (Zoom) -- Bo Li (UIUC) -- Secure
 Learning in Adversarial Environments -- AI Seminar sponsored by
 Fortive
In-Reply-To: <CAJA0PPDBYj-1cMXfzFk36k2FZKONbena9-bcOKtSGFVJsuHtbQ@mail.gmail.com>
References: <CAJA0PPDBYj-1cMXfzFk36k2FZKONbena9-bcOKtSGFVJsuHtbQ@mail.gmail.com>
Message-ID: <CAJA0PPA6rkz7f5xhpWE4PRDZDvfCxcSPuKVHpHnMuCnuUeEvPQ@mail.gmail.com>

Dear all,

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow *1**2pm-1pm*:
https://cmu.zoom.us/j/92752459790?pwd=aWN0NXhxaitPTHlnaEZpYUFuNUk1UT09.

Bo Li (UIUC) (see details below) will be talking about adversarial attacks
and security problems in modern machine learning.

Thanks,
Shaojie

On Tue, Apr 20, 2021 at 1:15 PM Shaojie Bai <shaojieb at andrew.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (4/27)* from *1**2:00-1:00 PM
> (U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
> by Fortive <https://careers.fortive.com/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
> <http://www.cs.cmu.edu/~aiseminar/>
>
> On 4/27, *Bo Li* (UIUC) will be giving a talk on "*Secure Learning in
> Adversarial Environments*".
>
> *Title*: Secure Learning in Adversarial Environments
>
> *Talk Abstract*: Advances in machine learning have led to rapid and
> widespread deployment of learning based inference and decision making for
> safety-critical applications, such as autonomous driving and security
> diagnostics. Current machine learning systems, however, assume that
> training and test data follow the same, or similar, distributions, and do
> not consider active adversaries manipulating either distribution. Recent
> work has demonstrated that motivated adversaries can circumvent anomaly
> detection or other machine learning models at test time through evasion
> attacks, or can inject well-crafted malicious instances into training data
> to induce errors in inference time through poisoning attacks. In this talk,
> I will describe my recent research about security and privacy problems in
> machine learning systems. In particular, I will introduce several
> adversarial attacks in different domains, and discuss potential defensive
> approaches and principles, including game theoretic based and knowledge
> enabled robust learning paradigms, towards developing practical robust
> learning systems with robustness guarantees.
>
> *Speaker Bio*: Dr. Bo Li is an assistant professor in the department of
> Computer Science at University of Illinois at Urbana-Champaign, and the
> recipient of the Symantec Research Labs Fellowship, Rising Stars, MIT
> Technology Review TR-35 award, Intel Rising Star award, Amazon Research
> Award, and best paper awards in several machine learning and security
> conferences. Previously she was a postdoctoral researcher in UC Berkeley.
> Her research focuses on both theoretical and practical aspects of security,
> machine learning, privacy, game theory, and adversarial machine learning.
> She has designed several robust learning algorithms, scalable frameworks
> for achieving robustness for a range of learning methods, and a privacy
> preserving data publishing system. Her work have been featured by major
> publications and media outlets such as Nature, Wired, Fortune, and New York
> Times.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/92752459790?pwd=aWN0NXhxaitPTHlnaEZpYUFuNUk1UT09
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210426/a79c2376/attachment.html>

From shaojieb at andrew.cmu.edu  Tue Apr 27 12:00:24 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Tue, 27 Apr 2021 12:00:24 -0400
Subject: [CMU AI Seminar] Apr 27 at 12pm (Zoom) -- Bo Li (UIUC) -- Secure
 Learning in Adversarial Environments -- AI Seminar sponsored by
 Fortive
In-Reply-To: <CAJA0PPA6rkz7f5xhpWE4PRDZDvfCxcSPuKVHpHnMuCnuUeEvPQ@mail.gmail.com>
References: <CAJA0PPDBYj-1cMXfzFk36k2FZKONbena9-bcOKtSGFVJsuHtbQ@mail.gmail.com>
 <CAJA0PPA6rkz7f5xhpWE4PRDZDvfCxcSPuKVHpHnMuCnuUeEvPQ@mail.gmail.com>
Message-ID: <CAJA0PPBtKWszK1w+-aTnHDw1e1iCHZu1vR7P3rGYhjvbogmPVQ@mail.gmail.com>

Hi everyone, this talk is happening in 5 minutes :-)

On Mon, Apr 26, 2021 at 3:46 PM Shaojie Bai <shaojieb at andrew.cmu.edu> wrote:

> Dear all,
>
> Just a reminder that the CMU AI Seminar
> <http://www.cs.cmu.edu/~aiseminar/> is tomorrow *1**2pm-1pm*:
> https://cmu.zoom.us/j/92752459790?pwd=aWN0NXhxaitPTHlnaEZpYUFuNUk1UT09.
>
> Bo Li (UIUC) (see details below) will be talking about adversarial attacks
> and security problems in modern machine learning.
>
> Thanks,
> Shaojie
>
> On Tue, Apr 20, 2021 at 1:15 PM Shaojie Bai <shaojieb at andrew.cmu.edu>
> wrote:
>
>> Dear all,
>>
>> We look forward to seeing you *next Tuesday (4/27)* from *1**2:00-1:00
>> PM (U.S. Eastern time)* for the next talk of our *CMU AI seminar*,
>> sponsored by Fortive <https://careers.fortive.com/>.
>>
>> To learn more about the seminar series or see the future schedule, please
>> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>> <http://www.cs.cmu.edu/~aiseminar/>
>>
>> On 4/27, *Bo Li* (UIUC) will be giving a talk on "*Secure Learning in
>> Adversarial Environments*".
>>
>> *Title*: Secure Learning in Adversarial Environments
>>
>> *Talk Abstract*: Advances in machine learning have led to rapid and
>> widespread deployment of learning based inference and decision making for
>> safety-critical applications, such as autonomous driving and security
>> diagnostics. Current machine learning systems, however, assume that
>> training and test data follow the same, or similar, distributions, and do
>> not consider active adversaries manipulating either distribution. Recent
>> work has demonstrated that motivated adversaries can circumvent anomaly
>> detection or other machine learning models at test time through evasion
>> attacks, or can inject well-crafted malicious instances into training data
>> to induce errors in inference time through poisoning attacks. In this talk,
>> I will describe my recent research about security and privacy problems in
>> machine learning systems. In particular, I will introduce several
>> adversarial attacks in different domains, and discuss potential defensive
>> approaches and principles, including game theoretic based and knowledge
>> enabled robust learning paradigms, towards developing practical robust
>> learning systems with robustness guarantees.
>>
>> *Speaker Bio*: Dr. Bo Li is an assistant professor in the department of
>> Computer Science at University of Illinois at Urbana-Champaign, and the
>> recipient of the Symantec Research Labs Fellowship, Rising Stars, MIT
>> Technology Review TR-35 award, Intel Rising Star award, Amazon Research
>> Award, and best paper awards in several machine learning and security
>> conferences. Previously she was a postdoctoral researcher in UC Berkeley.
>> Her research focuses on both theoretical and practical aspects of security,
>> machine learning, privacy, game theory, and adversarial machine learning.
>> She has designed several robust learning algorithms, scalable frameworks
>> for achieving robustness for a range of learning methods, and a privacy
>> preserving data publishing system. Her work have been featured by major
>> publications and media outlets such as Nature, Wired, Fortune, and New York
>> Times.
>>
>> *Zoom Link*:
>> https://cmu.zoom.us/j/92752459790?pwd=aWN0NXhxaitPTHlnaEZpYUFuNUk1UT09
>>
>> Thanks,
>> Shaojie Bai (MLD)
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210427/7dfbe803/attachment.html>

From shaojieb at andrew.cmu.edu  Thu Apr 29 12:30:49 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Thu, 29 Apr 2021 12:30:49 -0400
Subject: [CMU AI Seminar] May 04 at 12pm (Zoom) -- Raquel Urtasun (U of
 Toronto) -- Next Generation Simulation for Self-driving Vehicles -- AI
 Seminar sponsored by Fortive
Message-ID: <CAJA0PPAp3gpccvyDf_BEvCRjjbCq=h=3_8FNRuNab0TkNF2S9w@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (5/4)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
by Fortive <https://careers.fortive.com/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
<http://www.cs.cmu.edu/~aiseminar/>

On 4/27, *Raquel Urtasun* (University of Toronto) will be giving a
talk on "*Next
Generation Simulation for Self-driving Vehicles*".

*Title*: Next Generation Simulation for Self-driving Vehicles

*Talk Abstract*: We are on the verge of a new era in which robotics and
artificial intelligence will play an important role in our daily lives.
Self-driving vehicles have the potential to redefine transportation as we
understand it today. Our roads will become safer and less congested, while
parking spots will be repurposed as leisure zones and parks. However, many
technological challenges remain as we pursue this future. In this talk I
will focus on simulation, which is key for both testing and training modern
autonomy systems.

*Speaker Bio*: Raquel Urtasun is a Professor in the Department of Computer
Science at the University of Toronto, a Canada Research Chair in Machine
Learning and Computer Vision and a co-founder of the Vector Institute for
AI. Prior to this she was the Chief Scientist of Uber ATG and the Head of
Uber ATG R&D. She received her Ph.D. from the Ecole Polytechnique Federal
de Lausanne (EPFL) in 2006 and did her postdoc at MIT and UC Berkeley. She
is a recipient of an NSERC EWR Steacie Award, an NVIDIA Pioneers of AI
Award, a Ministry of Education and Innovation Early Researcher Award, three
Google Faculty Research Awards, an Amazon Faculty Research Award, a
Connaught New Researcher Award, a Fallona Family Research Award and two
Best Paper Runner up Prize awarded CVPR in 2013 and 2017. She was also
named Chatelaine 2018 Woman of the year, and 2018 Toronto's top influencers
by Adweek magazine.

*Zoom Link*:
https://cmu.zoom.us/j/96036411310?pwd=M0tqYW1NamE5Z1hWeWpxUGVBWTZGZz09

Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210429/3a9a9406/attachment.html>

From shaojieb at andrew.cmu.edu  Mon May  3 12:46:40 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Mon, 3 May 2021 12:46:40 -0400
Subject: [CMU AI Seminar] May 04 at 12pm (Zoom) -- Raquel Urtasun (U of
 Toronto) -- Next Generation Simulation for Self-driving Vehicles -- AI
 Seminar sponsored by Fortive
In-Reply-To: <CAJA0PPAp3gpccvyDf_BEvCRjjbCq=h=3_8FNRuNab0TkNF2S9w@mail.gmail.com>
References: <CAJA0PPAp3gpccvyDf_BEvCRjjbCq=h=3_8FNRuNab0TkNF2S9w@mail.gmail.com>
Message-ID: <CAJA0PPB7=+7Y+4EaKEgRUnMuM3J0CJ+apkSqoxfQCuB5HbVP-w@mail.gmail.com>

Dear all,

Just a reminder that Raquel's talk (see information below) at the CMU AI
Seminar <http://www.cs.cmu.edu/~aiseminar/> is tomorrow *1**2pm-1pm*:
https://cmu.zoom.us/j/96036411310?pwd=M0tqYW1NamE5Z1hWeWpxUGVBWTZGZz09.

Thanks,
Shaojie

On Thu, Apr 29, 2021 at 12:30 PM Shaojie Bai <shaojieb at andrew.cmu.edu>
wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (5/4)* from *1**2:00-1:00 PM
> (U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
> by Fortive <https://careers.fortive.com/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
> <http://www.cs.cmu.edu/~aiseminar/>
>
> On 4/27, *Raquel Urtasun* (University of Toronto) will be giving a talk
> on "*Next Generation Simulation for Self-driving Vehicles*".
>
> *Title*: Next Generation Simulation for Self-driving Vehicles
>
> *Talk Abstract*: We are on the verge of a new era in which robotics and
> artificial intelligence will play an important role in our daily lives.
> Self-driving vehicles have the potential to redefine transportation as we
> understand it today. Our roads will become safer and less congested, while
> parking spots will be repurposed as leisure zones and parks. However, many
> technological challenges remain as we pursue this future. In this talk I
> will focus on simulation, which is key for both testing and training modern
> autonomy systems.
>
> *Speaker Bio*: Raquel Urtasun is a Professor in the Department of
> Computer Science at the University of Toronto, a Canada Research Chair in
> Machine Learning and Computer Vision and a co-founder of the Vector
> Institute for AI. Prior to this she was the Chief Scientist of Uber ATG and
> the Head of Uber ATG R&D. She received her Ph.D. from the Ecole
> Polytechnique Federal de Lausanne (EPFL) in 2006 and did her postdoc at MIT
> and UC Berkeley. She is a recipient of an NSERC EWR Steacie Award, an
> NVIDIA Pioneers of AI Award, a Ministry of Education and Innovation Early
> Researcher Award, three Google Faculty Research Awards, an Amazon Faculty
> Research Award, a Connaught New Researcher Award, a Fallona Family Research
> Award and two Best Paper Runner up Prize awarded CVPR in 2013 and 2017. She
> was also named Chatelaine 2018 Woman of the year, and 2018 Toronto's top
> influencers by Adweek magazine.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/96036411310?pwd=M0tqYW1NamE5Z1hWeWpxUGVBWTZGZz09
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210503/b23c9400/attachment.html>

From shaojieb at andrew.cmu.edu  Tue May  4 11:55:39 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Tue, 4 May 2021 11:55:39 -0400
Subject: [CMU AI Seminar] May 04 at 12pm (Zoom) -- Raquel Urtasun (U of
 Toronto) -- Next Generation Simulation for Self-driving Vehicles -- AI
 Seminar sponsored by Fortive
In-Reply-To: <CAJA0PPAp3gpccvyDf_BEvCRjjbCq=h=3_8FNRuNab0TkNF2S9w@mail.gmail.com>
References: <CAJA0PPAp3gpccvyDf_BEvCRjjbCq=h=3_8FNRuNab0TkNF2S9w@mail.gmail.com>
Message-ID: <CAJA0PPCkGqkMoqpCBE-vG0qT4mGiYbeDGX34OEs3CO3h7RUKWg@mail.gmail.com>

Reminder: Raquel's talk will start in <10 minutes!

On Thu, Apr 29, 2021 at 12:30 PM Shaojie Bai <shaojieb at andrew.cmu.edu>
wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (5/4)* from *1**2:00-1:00 PM
> (U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
> by Fortive <https://careers.fortive.com/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
> <http://www.cs.cmu.edu/~aiseminar/>
>
> On 4/27, *Raquel Urtasun* (University of Toronto) will be giving a talk
> on "*Next Generation Simulation for Self-driving Vehicles*".
>
> *Title*: Next Generation Simulation for Self-driving Vehicles
>
> *Talk Abstract*: We are on the verge of a new era in which robotics and
> artificial intelligence will play an important role in our daily lives.
> Self-driving vehicles have the potential to redefine transportation as we
> understand it today. Our roads will become safer and less congested, while
> parking spots will be repurposed as leisure zones and parks. However, many
> technological challenges remain as we pursue this future. In this talk I
> will focus on simulation, which is key for both testing and training modern
> autonomy systems.
>
> *Speaker Bio*: Raquel Urtasun is a Professor in the Department of
> Computer Science at the University of Toronto, a Canada Research Chair in
> Machine Learning and Computer Vision and a co-founder of the Vector
> Institute for AI. Prior to this she was the Chief Scientist of Uber ATG and
> the Head of Uber ATG R&D. She received her Ph.D. from the Ecole
> Polytechnique Federal de Lausanne (EPFL) in 2006 and did her postdoc at MIT
> and UC Berkeley. She is a recipient of an NSERC EWR Steacie Award, an
> NVIDIA Pioneers of AI Award, a Ministry of Education and Innovation Early
> Researcher Award, three Google Faculty Research Awards, an Amazon Faculty
> Research Award, a Connaught New Researcher Award, a Fallona Family Research
> Award and two Best Paper Runner up Prize awarded CVPR in 2013 and 2017. She
> was also named Chatelaine 2018 Woman of the year, and 2018 Toronto's top
> influencers by Adweek magazine.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/96036411310?pwd=M0tqYW1NamE5Z1hWeWpxUGVBWTZGZz09
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210504/7865c271/attachment.html>

From shaojieb at andrew.cmu.edu  Thu May  6 18:36:40 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Thu, 6 May 2021 18:36:40 -0400
Subject: [CMU AI Seminar] May 11 at 12pm (Zoom) -- Jon Kleinberg (Cornell) --
 Aligning Superhuman AI with Human Behavior: Chess as a Model System
 -- AI Seminar sponsored by Fortive
Message-ID: <CAJA0PPCRFysZYJgMqJN89H=pceEeSWibFrfWqXNhb0js_91CXw@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (5/11)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
by Fortive <https://careers.fortive.com/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
<http://www.cs.cmu.edu/~aiseminar/>

On 5/11, the Winner of 2007 ACM Prize in Computing *Jon Kleinberg* (Cornell
University) will be giving a talk on "*Aligning Superhuman AI with Human
Behavior: Chess as a Model System*".

*Title*: Aligning Superhuman AI with Human Behavior: Chess as a Model System

*Talk Abstract*: In domains where AI systems have achieved superhuman
performance, there is an opportunity to study the similarities and
contrasts between human and AI behaviors at the level of granular actions,
not just aggregate performance. Such an analysis can yield several
potential sources of insight. First, by studying expert-level human
performance through the lens of systems that far surpass this performance,
we can try to characterize the settings in which human errors are most
likely to occur. Second, we can try to adapt high-performing AI systems to
match human behavior as closely as possible at an action-by-action level,

We pursue these goals in a domain with a long history in AI: chess. For our
purposes, chess provides a domain with many different levels of human
expertise, together with data from hundreds of millions of online games
that each record a specific decision together with its context. However,
applying existing chess engines to this data, including an open-source
implementation of AlphaZero, we find that they do not predict human moves
well.

We develop new methods for predicting human decisions at a move-by-move
level much more accurately than existing engines, and in a way that is
tunable to fine-grained differences in human skill. From this, we discover
aspects of chess positions that serve as predictors of human error, as well
as algorithms that are able to operate in this domain in a more
"human-like" way. One of our algorithms, the Maia chess engine, can be
tried at lichess.org (https://lichess.org/@/maia1), where it has played
over 300,000 games to date with users of the platform.

*Speaker Bio*: Jon Kleinberg is the Tisch University Professor in the
Departments of Computer Science and Information Science at Cornell
University. His research focuses on the interaction of algorithms and
networks, the roles they play in large-scale social and information
systems, and their broader societal implications. He is a member of the
National Academy of Sciences and the National Academy of Engineering, and
the recipient of MacArthur, Packard, Simons, Sloan, and Vannevar Bush
research fellowships, as well awards including the Harvey Prize, the
Nevanlinna Prize, and the ACM Prize in Computing.

*Zoom Link*:
https://cmu.zoom.us/j/98138716931?pwd=TmUyWjlTSnZsbEtvTjd3OGVXUXNzZz09
<https://www.google.com/url?q=https://cmu.zoom.us/j/98138716931?pwd%3DTmUyWjlTSnZsbEtvTjd3OGVXUXNzZz09&sa=D&source=calendar&ust=1620620634218000&usg=AOvVaw0uYZqBAG-_e0ZznJNuLFJl>


Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210506/e8ff55d0/attachment.html>

From shaojieb at andrew.cmu.edu  Mon May 10 13:02:58 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Mon, 10 May 2021 13:02:58 -0400
Subject: [CMU AI Seminar] May 11 at 12pm (Zoom) -- Jon Kleinberg (Cornell)
 -- Aligning Superhuman AI with Human Behavior: Chess as a Model
 System -- AI Seminar sponsored by Fortive
In-Reply-To: <CAJA0PPCRFysZYJgMqJN89H=pceEeSWibFrfWqXNhb0js_91CXw@mail.gmail.com>
References: <CAJA0PPCRFysZYJgMqJN89H=pceEeSWibFrfWqXNhb0js_91CXw@mail.gmail.com>
Message-ID: <CAJA0PPA0P7VgoO6Wn10Yr4nRWt4Z2TNfFjg2R3qCB4-0E7OYdA@mail.gmail.com>

Dear all,

Just a reminder that we are meeting tomorrow (*5/11 Tuesday*) at *12pm* for
the CMU AI Seminar with *Jon Kleinberg (Cornell)*.

Jon will share (and is very excited about) some of his latest research on
using chess as a model system to study algorithmic and human behavior for
problem-solving. See more information about the talk at:
http://www.cs.cmu.edu/~aiseminar/abstract/21-05-11.html

*Zoom Link*:
https://cmu.zoom.us/j/98138716931?pwd=TmUyWjlTSnZsbEtvTjd3OGVXUXNzZz09
<https://www.google.com/url?q=https://cmu.zoom.us/j/98138716931?pwd%3DTmUyWjlTSnZsbEtvTjd3OGVXUXNzZz09&sa=D&source=calendar&ust=1620620634218000&usg=AOvVaw0uYZqBAG-_e0ZznJNuLFJl>


Best,
Shaojie

On Thu, May 6, 2021 at 6:36 PM Shaojie Bai <shaojieb at andrew.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (5/11)* from *1**2:00-1:00 PM
> (U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
> by Fortive <https://careers.fortive.com/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
> <http://www.cs.cmu.edu/~aiseminar/>
>
> On 5/11, the Winner of 2007 ACM Prize in Computing *Jon Kleinberg* (Cornell
> University) will be giving a talk on "*Aligning Superhuman AI with Human
> Behavior: Chess as a Model System*".
>
> *Title*: Aligning Superhuman AI with Human Behavior: Chess as a Model
> System
>
> *Talk Abstract*: In domains where AI systems have achieved superhuman
> performance, there is an opportunity to study the similarities and
> contrasts between human and AI behaviors at the level of granular actions,
> not just aggregate performance. Such an analysis can yield several
> potential sources of insight. First, by studying expert-level human
> performance through the lens of systems that far surpass this performance,
> we can try to characterize the settings in which human errors are most
> likely to occur. Second, we can try to adapt high-performing AI systems to
> match human behavior as closely as possible at an action-by-action level,
>
> We pursue these goals in a domain with a long history in AI: chess. For
> our purposes, chess provides a domain with many different levels of human
> expertise, together with data from hundreds of millions of online games
> that each record a specific decision together with its context. However,
> applying existing chess engines to this data, including an open-source
> implementation of AlphaZero, we find that they do not predict human moves
> well.
>
> We develop new methods for predicting human decisions at a move-by-move
> level much more accurately than existing engines, and in a way that is
> tunable to fine-grained differences in human skill. From this, we discover
> aspects of chess positions that serve as predictors of human error, as well
> as algorithms that are able to operate in this domain in a more
> "human-like" way. One of our algorithms, the Maia chess engine, can be
> tried at lichess.org (https://lichess.org/@/maia1), where it has played
> over 300,000 games to date with users of the platform.
>
> *Speaker Bio*: Jon Kleinberg is the Tisch University Professor in the
> Departments of Computer Science and Information Science at Cornell
> University. His research focuses on the interaction of algorithms and
> networks, the roles they play in large-scale social and information
> systems, and their broader societal implications. He is a member of the
> National Academy of Sciences and the National Academy of Engineering, and
> the recipient of MacArthur, Packard, Simons, Sloan, and Vannevar Bush
> research fellowships, as well awards including the Harvey Prize, the
> Nevanlinna Prize, and the ACM Prize in Computing.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/98138716931?pwd=TmUyWjlTSnZsbEtvTjd3OGVXUXNzZz09
> <https://www.google.com/url?q=https://cmu.zoom.us/j/98138716931?pwd%3DTmUyWjlTSnZsbEtvTjd3OGVXUXNzZz09&sa=D&source=calendar&ust=1620620634218000&usg=AOvVaw0uYZqBAG-_e0ZznJNuLFJl>
>
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210510/85b3bc0e/attachment.html>

From shaojieb at andrew.cmu.edu  Sat May 15 15:15:31 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Sat, 15 May 2021 15:15:31 -0400
Subject: [CMU AI Seminar] May 18 at 12pm (Zoom) -- Vladlen Koltun (Intel Labs)
 -- Are we publishing too much? -- AI Seminar sponsored by Fortive
Message-ID: <CAJA0PPCB-n5ykmbsUh7d1uSaW60GBNB+g0QP2F0Fh8DGBi5AFQ@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (5/18)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
by Fortive <https://careers.fortive.com/>.

This is the last AI seminar talk of the Spring 2021 semester :-)

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
<http://www.cs.cmu.edu/~aiseminar/>

On 5/18, *Vladlen Koltun* (Intel Labs) will be giving a talk on "*Are we
publishing too much?*".

*Title*:  Are we publishing too much?

*Related paper(s)*:  https://arxiv.org/abs/2102.03234 (+ one coming out
soon)

*Speaker Bio*:  Vladlen Koltun is the Chief Scientist for Intelligent
Systems at Intel. He directs the Intelligent Systems Lab, which conducts
high-impact basic research in computer vision, machine learning, robotics,
and related areas. He has mentored more than 50 PhD students, postdocs,
research scientists, and PhD student interns, many of whom are now
successful research leaders.

*Zoom Link*:  https://cmu.zoom.us/j/93527537239?pwd=SStuS0U3NHZpd0NORVMwWGI2Sk1TQT09
<https://cmu.zoom.us/j/93527537239?pwd=SStuS0U3NHZpd0NORVMwWGI2Sk1TQT09>

Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210515/7ca495a9/attachment.html>

From shaojieb at andrew.cmu.edu  Mon May 17 15:26:02 2021
From: shaojieb at andrew.cmu.edu (Shaojie Bai)
Date: Mon, 17 May 2021 15:26:02 -0400
Subject: [CMU AI Seminar] May 18 at 12pm (Zoom) -- Vladlen Koltun (Intel
 Labs) -- Are we publishing too much? -- AI Seminar sponsored by Fortive
In-Reply-To: <CAJA0PPCB-n5ykmbsUh7d1uSaW60GBNB+g0QP2F0Fh8DGBi5AFQ@mail.gmail.com>
References: <CAJA0PPCB-n5ykmbsUh7d1uSaW60GBNB+g0QP2F0Fh8DGBi5AFQ@mail.gmail.com>
Message-ID: <CAJA0PPBFYO5VKKp=1fqq0AFrgEpKAUGjE-uO9up8Dh=BbZBs2Q@mail.gmail.com>

Dear all,

A reminder that we are meeting tomorrow (*5/18 Tuesday*) at *12pm* for the
CMU AI Seminar with *Vladlen Koltun (Intel Labs)*.

Vladlen will talk about his recent research on evaluating researchers and why
h-index may no longer be a good metric <https://arxiv.org/abs/2102.03234>.
See more information about the talk at:
http://www.cs.cmu.edu/~aiseminar/abstract/21-05-18.html
<http://www.cs.cmu.edu/~aiseminar/abstract/21-05-11.html>

*Zoom Link*:
https://cmu.zoom.us/j/93527537239?pwd=SStuS0U3NHZpd0NORVMwWGI2Sk1TQT09

Best,
Shaojie

On Sat, May 15, 2021 at 3:15 PM Shaojie Bai <shaojieb at andrew.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (5/18)* from *1**2:00-1:00 PM
> (U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
> by Fortive <https://careers.fortive.com/>.
>
> This is the last AI seminar talk of the Spring 2021 semester :-)
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
> <http://www.cs.cmu.edu/~aiseminar/>
>
> On 5/18, *Vladlen Koltun* (Intel Labs) will be giving a talk on "*Are we
> publishing too much?*".
>
> *Title*:  Are we publishing too much?
>
> *Related paper(s)*:  https://arxiv.org/abs/2102.03234 (+ one coming out
> soon)
>
> *Speaker Bio*:  Vladlen Koltun is the Chief Scientist for Intelligent
> Systems at Intel. He directs the Intelligent Systems Lab, which conducts
> high-impact basic research in computer vision, machine learning, robotics,
> and related areas. He has mentored more than 50 PhD students, postdocs,
> research scientists, and PhD student interns, many of whom are now
> successful research leaders.
>
> *Zoom Link*:  https://cmu.zoom.us/j/93527537239?pwd=SStuS0U3NHZpd0NORVMwWGI2Sk1TQT09
> <https://cmu.zoom.us/j/93527537239?pwd=SStuS0U3NHZpd0NORVMwWGI2Sk1TQT09>
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210517/047a0a7c/attachment.html>

From shaojieb at cs.cmu.edu  Tue Sep 14 13:30:57 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Tue, 14 Sep 2021 13:30:57 -0400
Subject: [CMU AI Seminar] Sep 21 at 12pm (Zoom) -- Stephan Hoyer (Google) --
 Accelerating computational fluid dynamics with deep learning -- AI Seminar
 sponsored by Morgan Stanley
Message-ID: <CAJA0PPA3VFPDA-U=1_US=emkQy5esSdNq51z-MqOw9KQBN3ZHA@mail.gmail.com>

Dear all,

Welcome to the CMU AI Seminar for the Fall 2021 semester! For the new
semester, we are excited to announce that *Morgan Stanley* has become the
new sponsor of the seminar series, which will start on September 21.

We look forward to seeing you *next Tuesday (9/21)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
by Morgan Stanley <https://www.morganstanley.com/about-us/technology/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.

On 9/21, *Stephan Hoyer* (Google) will be giving a talk on "*Accelerating
Computational Fluid Dynamics with Deep Learning*" and share some of
the exciting progress in deep learning for scientific computing.

*Title*: Accelerating Computational Fluid Dynamics with Deep Learning

*Talk Abstract*: How can machine learning help large-scale scientific
simulation? Accurate simulation of fluids is important for problems like
engineering design and climate modeling, but is very computationally
demanding. In this talk, I'll give an overview of a line of research at
Google, where we've been using end-to-end deep learning to improve
approximations inside traditional numerical solvers. For 2D turbulent
flows, our models are up to two orders of magnitude faster than traditional
solvers with the same accuracy on the same hardware, and can still
generalize to very different types of flows from those on which they were
trained.

*Speaker Bio*:  Stephan Hoyer is a staff engineer at Google Research. He
works on deep learning for science, with a focus on physical simulations
and applications in climate/weather modeling. His research centers on the
hypothesis that automatic differentiation software, hardware accelerators
and deep learning are poised to transform traditional scientific computing,
by vastly accelerating and improving existing numerical models. He also
frequently contributes to open source tools for scientific computing in
Python, including JAX and NumPy. Before Google, he was a data scientist at
The Climate Corporation, and received his Ph.D in physics from UC Berkeley.

*Zoom Link*:
https://cmu.zoom.us/j/99723728452?pwd=YnBOa0ZDSXRxRmdSYTJiNGNVVFJ4UT09

Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210914/de0862ca/attachment.html>

From shaojieb at cs.cmu.edu  Mon Sep 20 11:33:36 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Mon, 20 Sep 2021 11:33:36 -0400
Subject: [CMU AI Seminar] Sep 21 at 12pm (Zoom) -- Stephan Hoyer (Google)
 -- Accelerating computational fluid dynamics with deep learning -- AI
 Seminar sponsored by Morgan Stanley
In-Reply-To: <CAJA0PPA3VFPDA-U=1_US=emkQy5esSdNq51z-MqOw9KQBN3ZHA@mail.gmail.com>
References: <CAJA0PPA3VFPDA-U=1_US=emkQy5esSdNq51z-MqOw9KQBN3ZHA@mail.gmail.com>
Message-ID: <CAJA0PPAXKauWtjofZ-vg5aZ6j04cnSX8=HvhnTRtQnw6k4t8bA@mail.gmail.com>

Hi all,

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow *12pm-1pm*:
https://cmu.zoom.us/j/99723728452?pwd=YnBOa0ZDSXRxRmdSYTJiNGNVVFJ4UT09.

*Stephan Hoyer (Google)* will be talking about some latest progress in
computational fluid dynamics, PDEs and deep learning! At the beginning of
the seminar tomorrow, the sponsor of our AI seminar (Morgan Stanley) will
also briefly introduce their AI team to kick off the fall 2021 series.

Thanks,
Shaojie

On Tue, Sep 14, 2021 at 1:30 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:

> Dear all,
>
> Welcome to the CMU AI Seminar for the Fall 2021 semester! For the new
> semester, we are excited to announce that *Morgan Stanley* has become the
> new sponsor of the seminar series, which will start on September 21.
>
> We look forward to seeing you *next Tuesday (9/21)* from *1**2:00-1:00 PM
> (U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
> by Morgan Stanley <https://www.morganstanley.com/about-us/technology/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>
> On 9/21, *Stephan Hoyer* (Google) will be giving a talk on "*Accelerating
> Computational Fluid Dynamics with Deep Learning*" and share some of
> the exciting progress in deep learning for scientific computing.
>
> *Title*: Accelerating Computational Fluid Dynamics with Deep Learning
>
> *Talk Abstract*: How can machine learning help large-scale scientific
> simulation? Accurate simulation of fluids is important for problems like
> engineering design and climate modeling, but is very computationally
> demanding. In this talk, I'll give an overview of a line of research at
> Google, where we've been using end-to-end deep learning to improve
> approximations inside traditional numerical solvers. For 2D turbulent
> flows, our models are up to two orders of magnitude faster than traditional
> solvers with the same accuracy on the same hardware, and can still
> generalize to very different types of flows from those on which they were
> trained.
>
> *Speaker Bio*:  Stephan Hoyer is a staff engineer at Google Research. He
> works on deep learning for science, with a focus on physical simulations
> and applications in climate/weather modeling. His research centers on the
> hypothesis that automatic differentiation software, hardware accelerators
> and deep learning are poised to transform traditional scientific computing,
> by vastly accelerating and improving existing numerical models. He also
> frequently contributes to open source tools for scientific computing in
> Python, including JAX and NumPy. Before Google, he was a data scientist at
> The Climate Corporation, and received his Ph.D in physics from UC Berkeley.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/99723728452?pwd=YnBOa0ZDSXRxRmdSYTJiNGNVVFJ4UT09
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210920/570f4ead/attachment.html>

From shaojieb at cs.cmu.edu  Fri Sep 24 17:42:02 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Fri, 24 Sep 2021 17:42:02 -0400
Subject: [CMU AI Seminar] Sep 28 at 12pm (Zoom) -- Ashique Khudabukhsh (CMU)
 -- Novel Frameworks for Quantifying Political Polarization and Mitigating
 Hate Speech -- AI Seminar sponsored by Morgan Stanley
Message-ID: <CAJA0PPDBYHKhRrdESa67POuR+J7a+=_DsxPprn4oeV5OMOt7Dw@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (9/28)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
by Morgan Stanley <https://www.morganstanley.com/about-us/technology/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.

On 9/28, *Ahique Khudabukhsh* (CMU LTI) will be giving a talk on "*Novel
Frameworks for Quantifying Political Polarization and Mitigating Hate
Speech*".

*Title*:  Novel Frameworks for Quantifying Political Polarization and
Mitigating Hate Speech

*Talk Abstract*: This talk is divided into two parts. Each part summarizes
a broad line of NLP research outlining a new framework. The first part of
the talk presents a new methodology that offers a fresh perspective on
interpreting and understanding political polarization through machine
translation. I begin with a novel proposition that two sub-communities
viewing different US cable news networks are speaking in two different
languages. Next, I demonstrate that with this assumption, modern machine
translation methods can provide a simple yet powerful and interpretable
framework to understand the differences between two (or more) large-scale
social media discussion data sets at the granularity of words.

The second part of the talk presents a new direction for mitigating online
hate. Much of the existing research geared toward making the internet a
safer place involves identifying hate speech as the first step. However,
little or no attention is given to the possibility that the not-hate-speech
subset of the corpus may contain content with potentially positive societal
impact. I introduce two new tasks, namely hope speech detection --
detecting hostility-diffusing, peace-seeking content -- and help speech
detection -- detecting content supportive of a disenfranchised minority. I
illustrate applications of these two new tasks in the context of the
most-recent India-Pakistan conflict triggered by the 2019 Pulwama terror
attack, and the longstanding Rohingya refugee crisis that rendered more
than 700,000 people homeless. Beyond the framework novelty of focusing on
the positive content, this work addresses several practical challenges that
arise from multilingual texts in a noisy, social media setting.

*Speaker Bio*:  Ashique Khudabukhsh is an assistant professor at the
Golisano College of Computing and Information Sciences, Rochester Institute
of Technology (RIT). His current research lies at the intersection of NLP
and AI for Social Impact as applied to: (i) globally important events
arising in linguistically diverse regions requiring methods to tackle
practical challenges involving multilingual, noisy, social media texts; and
(ii) polarization in the context of the current US political crisis. In
addition to having his research been accepted at top artificial
intelligence conferences and journals, his work has also received
widespread international media attention that includes multiple coverage
from BBC, Wired, Salon, The Independent, VentureBeat, and Digital Trends.
Prior to joining RIT, Ashique was a Project Scientist at the Language
Technologies Institute, Carnegie Mellon University (CMU) mentored by Prof.
Tom Mitchell. Prior to this, he was a postdoc mentored by Prof. Jaime
Carbonell at CMU. His PhD thesis (Computer Science Department, CMU, also
advised by Prof. Jaime Carbonell) focused on distributed active learning.

*Zoom Link*:
https://cmu.zoom.us/j/96000432347?pwd=TXVhU2dlSVZyM3hjTzVVVEhUclVIdz09

Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210924/e96964b7/attachment.html>

From shaojieb at cs.cmu.edu  Mon Sep 27 15:49:52 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Mon, 27 Sep 2021 15:49:52 -0400
Subject: [CMU AI Seminar] Sep 28 at 12pm (Zoom) -- Ashique Khudabukhsh
 (CMU) -- Novel Frameworks for Quantifying Political Polarization and
 Mitigating Hate Speech -- AI Seminar sponsored by Morgan Stanley
In-Reply-To: <CAJA0PPDBYHKhRrdESa67POuR+J7a+=_DsxPprn4oeV5OMOt7Dw@mail.gmail.com>
References: <CAJA0PPDBYHKhRrdESa67POuR+J7a+=_DsxPprn4oeV5OMOt7Dw@mail.gmail.com>
Message-ID: <CAJA0PPCne_L4wKcwu7QPNAGhF5Rsfj+QD_tqNG-+XCEoXEj=Rg@mail.gmail.com>

Hi all,

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow *12pm-1pm*:
https://cmu.zoom.us/j/96000432347?pwd=TXVhU2dlSVZyM3hjTzVVVEhUclVIdz09.

*Ashique Khudabukhsh (CMU/RIT)* will be talking about his research on
quantifying political polarization and mitigating online hate.

Thanks,
Shaojie

On Fri, Sep 24, 2021 at 5:42 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (9/28)* from *1**2:00-1:00 PM
> (U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
> by Morgan Stanley <https://www.morganstanley.com/about-us/technology/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>
> On 9/28, *Ahique Khudabukhsh* (CMU LTI) will be giving a talk on "*Novel
> Frameworks for Quantifying Political Polarization and Mitigating Hate
> Speech*".
>
> *Title*:  Novel Frameworks for Quantifying Political Polarization and
> Mitigating Hate Speech
>
> *Talk Abstract*: This talk is divided into two parts. Each part
> summarizes a broad line of NLP research outlining a new framework. The
> first part of the talk presents a new methodology that offers a fresh
> perspective on interpreting and understanding political polarization
> through machine translation. I begin with a novel proposition that two
> sub-communities viewing different US cable news networks are speaking in
> two different languages. Next, I demonstrate that with this assumption,
> modern machine translation methods can provide a simple yet powerful and
> interpretable framework to understand the differences between two (or more)
> large-scale social media discussion data sets at the granularity of words.
>
> The second part of the talk presents a new direction for mitigating online
> hate. Much of the existing research geared toward making the internet a
> safer place involves identifying hate speech as the first step. However,
> little or no attention is given to the possibility that the not-hate-speech
> subset of the corpus may contain content with potentially positive societal
> impact. I introduce two new tasks, namely hope speech detection --
> detecting hostility-diffusing, peace-seeking content -- and help speech
> detection -- detecting content supportive of a disenfranchised minority. I
> illustrate applications of these two new tasks in the context of the
> most-recent India-Pakistan conflict triggered by the 2019 Pulwama terror
> attack, and the longstanding Rohingya refugee crisis that rendered more
> than 700,000 people homeless. Beyond the framework novelty of focusing on
> the positive content, this work addresses several practical challenges that
> arise from multilingual texts in a noisy, social media setting.
>
> *Speaker Bio*:  Ashique Khudabukhsh is an assistant professor at the
> Golisano College of Computing and Information Sciences, Rochester Institute
> of Technology (RIT). His current research lies at the intersection of NLP
> and AI for Social Impact as applied to: (i) globally important events
> arising in linguistically diverse regions requiring methods to tackle
> practical challenges involving multilingual, noisy, social media texts; and
> (ii) polarization in the context of the current US political crisis. In
> addition to having his research been accepted at top artificial
> intelligence conferences and journals, his work has also received
> widespread international media attention that includes multiple coverage
> from BBC, Wired, Salon, The Independent, VentureBeat, and Digital Trends.
> Prior to joining RIT, Ashique was a Project Scientist at the Language
> Technologies Institute, Carnegie Mellon University (CMU) mentored by Prof.
> Tom Mitchell. Prior to this, he was a postdoc mentored by Prof. Jaime
> Carbonell at CMU. His PhD thesis (Computer Science Department, CMU, also
> advised by Prof. Jaime Carbonell) focused on distributed active learning.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/96000432347?pwd=TXVhU2dlSVZyM3hjTzVVVEhUclVIdz09
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20210927/59ec66ca/attachment.html>

From volks at albrecht-lorenz.de  Thu Oct  7 01:56:03 2021
From: volks at albrecht-lorenz.de (Trust Wallet)
Date: Thu, 7 Oct 2021 07:56:03 +0200
Subject: Account_Notification
Message-ID: <B915C2AA75734F79ADAAF56CF74A6FBB.MAI@home>

?*Nur in der App. Nur solange der Vorrat reicht. Gilt f?r das ausgew?hlte Sortiment. Nicht mit anderen Aktionscodes kombinierbar. Nach Codeeingabe werden dir 20% im Bestellprozess abgezogen. G?ltig am 06.10.2021.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211007/4ccf8d7d/attachment.html>

From shaojieb at cs.cmu.edu  Sat Oct  9 12:29:07 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Sat, 9 Oct 2021 12:29:07 -0400
Subject: [CMU AI Seminar] Oct 12 at 12pm (Zoom) -- Kashif Rasul (Morgan
 Stanley) -- Modern Neural Probabilistic Forecasting -- AI Seminar sponsored
 by Morgan Stanley
Message-ID: <CAJA0PPA5PmN-3QBsp3ATXPe5ezO+KfPsRv2k7CX1xC3X6O_Z8Q@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (10/12)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
by Morgan Stanley <https://www.morganstanley.com/about-us/technology/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.

On 10/12, *Kashif Rasul* (Morgan Stanley) will be giving a talk on "*Modern
Neural Probabilistic Forecasting*" and some of the interesting projects
that Morgan Stanley's ML research team has been working on.

*Title*:  Modern Neural Probabilistic Forecasting

*Talk Abstract*: In this talk we will introduce the foundations of Neural
Probabilistic Forecasting by reviewing its building blocks as well as some
issues. We will then introduce how probabilistic forecasting works in this
setting and talk about two recent papers which use Normalizing Flows and
Energy-based methods to tackle this problem.

*Speaker Bio*:  Kashif Rasul is a Machine Learning Research scientist
working on deep learning based methods and their applications at Morgan
Stanley.

*Zoom Link*:
https://cmu.zoom.us/j/91578896206?pwd=SVhjS05Nb0FkTllOMzROTElCdUVXdz09
<https://www.google.com/url?q=https://cmu.zoom.us/j/91578896206?pwd%3DSVhjS05Nb0FkTllOMzROTElCdUVXdz09&sa=D&source=calendar&ust=1634228491812285&usg=AOvVaw2XUhPjMASH1UOInRbqmS9W>

Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211009/c2dbaf99/attachment.html>

From shaojieb at cs.cmu.edu  Mon Oct 11 16:58:36 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Mon, 11 Oct 2021 16:58:36 -0400
Subject: [CMU AI Seminar] Oct 12 at 12pm (Zoom) -- Kashif Rasul (Morgan
 Stanley) -- Modern Neural Probabilistic Forecasting -- AI Seminar sponsored
 by Morgan Stanley
In-Reply-To: <CAJA0PPA5PmN-3QBsp3ATXPe5ezO+KfPsRv2k7CX1xC3X6O_Z8Q@mail.gmail.com>
References: <CAJA0PPA5PmN-3QBsp3ATXPe5ezO+KfPsRv2k7CX1xC3X6O_Z8Q@mail.gmail.com>
Message-ID: <CAJA0PPDFHiNxULR-67hbJC-SaTLcvYYsW5cyOb8Qb0egof=pPA@mail.gmail.com>

Hi all,

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow *12pm-1pm*:
https://cmu.zoom.us/j/91578896206?pwd=SVhjS05Nb0FkTllOMzROTElCdUVXdz09.

*Kashif Rasul (Morgan Stanley)* will be talking about his research on
neural probabilistic forecasting.

Thanks,
Shaojie

On Sat, Oct 9, 2021 at 12:29 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (10/12)* from *1**2:00-1:00
> PM (U.S. Eastern time)* for the next talk of our *CMU AI seminar*,
> sponsored by Morgan Stanley
> <https://www.morganstanley.com/about-us/technology/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>
> On 10/12, *Kashif Rasul* (Morgan Stanley) will be giving a talk on "*Modern
> Neural Probabilistic Forecasting*" and some of the interesting projects
> that Morgan Stanley's ML research team has been working on.
>
> *Title*:  Modern Neural Probabilistic Forecasting
>
> *Talk Abstract*: In this talk we will introduce the foundations of Neural
> Probabilistic Forecasting by reviewing its building blocks as well as some
> issues. We will then introduce how probabilistic forecasting works in this
> setting and talk about two recent papers which use Normalizing Flows and
> Energy-based methods to tackle this problem.
>
> *Speaker Bio*:  Kashif Rasul is a Machine Learning Research scientist
> working on deep learning based methods and their applications at Morgan
> Stanley.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/91578896206?pwd=SVhjS05Nb0FkTllOMzROTElCdUVXdz09
> <https://www.google.com/url?q=https://cmu.zoom.us/j/91578896206?pwd%3DSVhjS05Nb0FkTllOMzROTElCdUVXdz09&sa=D&source=calendar&ust=1634228491812285&usg=AOvVaw2XUhPjMASH1UOInRbqmS9W>
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211011/32b57220/attachment.html>

From shaojieb at cs.cmu.edu  Thu Oct 14 17:41:47 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Thu, 14 Oct 2021 17:41:47 -0400
Subject: [CMU AI Seminar] Oct 19 at 12pm (Zoom) -- Stefano Ermon (Stanford) --
 Generative Modeling by Estimating Gradients of the Data Distribution
 -- AI Seminar sponsored by Morgan Stanley
Message-ID: <CAJA0PPDqfj+B4HmOnt9q5xdWm9ACEnxJUx-KupQc0qCCyQWJ7w@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (10/19)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
by Morgan Stanley <https://www.morganstanley.com/about-us/technology/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.

On 10/19, *Stefano Ermon* (Stanford) will be giving a talk on "*Generative
Modeling by Estimating Gradients of the Data Distribution*" and their
research on score-based models.

*Title*: Generative Modeling by Estimating Gradients of the Data
Distribution

*Talk Abstract*: Existing generative models are typically based on explicit
representations of probability distributions (e.g., autoregressive or VAEs)
or implicit sampling procedures (e.g., GANs). We propose an alternative
approach based on modeling directly the vector field of gradients of the
data distribution (scores). Our framework allows flexible architectures,
requires no sampling during training or the use of adversarial training
methods. Additionally, score-based generative models enable exact
likelihood evaluation through connections with normalizing flows. We
produce samples comparable to GANs, achieving new state-of-the-art
inception scores, and competitive likelihoods on image datasets.

*Speaker Bio*:  Stefano Ermon is an Assistant Professor of Computer Science
in the CS Department at Stanford University, where he is affiliated with
the Artificial Intelligence Laboratory, and a fellow of the Woods Institute
for the Environment. His research is centered on techniques for
probabilistic modeling of data and is motivated by applications in the
emerging field of computational sustainability. He has won several awards,
including Best Paper Awards (ICLR, AAAI, UAI and CP), a NSF Career Award,
ONR and AFOSR Young Investigator Awards, a Sony Faculty Innovation Award, a
Hellman Faculty Fellowship, Microsoft Research Fellowship, Sloan
Fellowship, and the IJCAI Computers and Thought Award. Stefano earned his
Ph.D. in Computer Science at Cornell University in 2015.

*Zoom Link*:
https://cmu.zoom.us/j/99631487756?pwd=a2NORWwwaWlybjNZemo3N2h1RkZ2dz09

Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211014/d4179739/attachment.html>

From shaojieb at cs.cmu.edu  Mon Oct 18 16:16:32 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Mon, 18 Oct 2021 16:16:32 -0400
Subject: [CMU AI Seminar] Oct 19 at 12pm (Zoom) -- Stefano Ermon
 (Stanford) -- Generative Modeling by Estimating Gradients of the Data
 Distribution -- AI Seminar sponsored by Morgan Stanley
In-Reply-To: <CAJA0PPDqfj+B4HmOnt9q5xdWm9ACEnxJUx-KupQc0qCCyQWJ7w@mail.gmail.com>
References: <CAJA0PPDqfj+B4HmOnt9q5xdWm9ACEnxJUx-KupQc0qCCyQWJ7w@mail.gmail.com>
Message-ID: <CAJA0PPC+ZQX3KW0suatpMLmDui2c+UdCa_C4QWEqAQ_2uAZ2DA@mail.gmail.com>

Hi all,

Just a reminder that the CMU AI Seminar is tomorrow 12pm-1pm:
https://cmu.zoom.us/j/99631487756?pwd=a2NORWwwaWlybjNZemo3N2h1RkZ2dz09  .

Stefano Ermon (Stanford) will be talking about his group's latest research
on generative modeling with score-based methods.

Thanks,
Shaojie

On Thu, Oct 14, 2021 at 5:41 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (10/19)* from *1**2:00-1:00
> PM (U.S. Eastern time)* for the next talk of our *CMU AI seminar*,
> sponsored by Morgan Stanley
> <https://www.morganstanley.com/about-us/technology/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>
> On 10/19, *Stefano Ermon* (Stanford) will be giving a talk on "*Generative
> Modeling by Estimating Gradients of the Data Distribution*" and their
> research on score-based models.
>
> *Title*: Generative Modeling by Estimating Gradients of the Data
> Distribution
>
> *Talk Abstract*: Existing generative models are typically based on
> explicit representations of probability distributions (e.g., autoregressive
> or VAEs) or implicit sampling procedures (e.g., GANs). We propose an
> alternative approach based on modeling directly the vector field of
> gradients of the data distribution (scores). Our framework allows flexible
> architectures, requires no sampling during training or the use of
> adversarial training methods. Additionally, score-based generative models
> enable exact likelihood evaluation through connections with normalizing
> flows. We produce samples comparable to GANs, achieving new
> state-of-the-art inception scores, and competitive likelihoods on image
> datasets.
>
> *Speaker Bio*:  Stefano Ermon is an Assistant Professor of Computer
> Science in the CS Department at Stanford University, where he is affiliated
> with the Artificial Intelligence Laboratory, and a fellow of the Woods
> Institute for the Environment. His research is centered on techniques for
> probabilistic modeling of data and is motivated by applications in the
> emerging field of computational sustainability. He has won several awards,
> including Best Paper Awards (ICLR, AAAI, UAI and CP), a NSF Career Award,
> ONR and AFOSR Young Investigator Awards, a Sony Faculty Innovation Award, a
> Hellman Faculty Fellowship, Microsoft Research Fellowship, Sloan
> Fellowship, and the IJCAI Computers and Thought Award. Stefano earned his
> Ph.D. in Computer Science at Cornell University in 2015.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/99631487756?pwd=a2NORWwwaWlybjNZemo3N2h1RkZ2dz09
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211018/4e60b414/attachment.html>

From nihars at cs.cmu.edu  Sat Oct 23 12:31:57 2021
From: nihars at cs.cmu.edu (Nihar Shah)
Date: Sat, 23 Oct 2021 12:31:57 -0400
Subject: Talk on Auctions and Prediction Markets for Scientific Peer Review
Message-ID: <CAF7+7kpgjBUvFW-qKujCiSfaEjE_AQwnUVd7RFLF5Q7uczZBuw@mail.gmail.com>

Dear all,

Siddarth Srinivasan and CMU alum Jamie Morgenstern from the University of
Washington will be talking about their recent work
<https://arxiv.org/pdf/2109.00923.pdf> on "Auctions and Prediction Markets
for Scientific Peer Review" in my group meeting over zoom. Might be of
interest to some of you -- please feel free to join if you are interested.

Date: Monday, October 25
Time: 2.30-3.30pm
Zoom link:

https://cmu.zoom.us/j/97934706422?pwd=UWN5WkczOUdWN1Zuc0lUeWxRY2ZlQT09

(please log in to zoom via CMU account)

Best,
Nihar
http://cs.cmu.edu/~nihars
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211023/c85f3bb3/attachment.html>

From shaojieb at cs.cmu.edu  Sat Oct 23 15:08:56 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Sat, 23 Oct 2021 15:08:56 -0400
Subject: Oct 26 at 12pm (Zoom) -- Chen Dan (CMU CSD) -- Sharp Statistical
 Guarantees for Adversarially Robust Gaussian Classification -- AI Seminar
 sponsored by Morgan Stanley
Message-ID: <CAJA0PPAeb_9y28LwVKY75NbO3-D0SJnu2U94uk-BxuY8_t3dyw@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (10/26)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI seminar*, sponsored
by Morgan Stanley <https://www.morganstanley.com/about-us/technology/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.

On 10/26, *Chen Dan* (CMU CSD) will be giving a talk on "*Sharp Statistical
Guarantees for Adversarially Robust Gaussian Classification*".

*Title*: Sharp Statistical Guarantees for Adversarially Robust Gaussian
Classification

*Talk Abstract*: Adversarial robustness has become a fundamental
requirement in modern machine learning applications. Yet, there has been
surprisingly little statistical understanding so far. In this work, we
provide the first result of the optimal minimax guarantees for the excess
risk for adversarially robust classification, under a Gaussian mixture
model studied by Schmidt et al. 2018. The results are stated in terms of
the Adversarial Signal-to-Noise Ratio (AdvSNR), which generalizes a similar
notion for standard linear classification to the adversarial setting. We
establish an excess risk lower bound and design a computationally efficient
estimator that achieves this optimal rate. Our results built upon a minimal
set of assumptions while covering a wide spectrum of adversarial
perturbations including L_p balls for any p>1. Joint work with Yuting Wei
and Pradeep Ravikumar.

*Speaker Bio*:  Chen Dan is a 6th year Ph.D. student at Computer Science
Department, Carnegie Mellon University, advised by Pradeep Ravikumar. His
research interest is in the broad area of robust statistical learning, with
an emphasis on the theoretical understanding and practical algorithms for
learning under various types of adversarial distribution shift. Prior to
joining CMU, Chen received his bachelor degree from School of EECS, Peking
University in 2016.

*Zoom Link*:
https://cmu.zoom.us/j/93155268338?pwd=VVZYTFFEMTNLZlJVY1NmU1c3cXUzZz09

Thanks,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211023/e6bb99f5/attachment.html>

From shaojieb at cs.cmu.edu  Mon Oct 25 20:18:10 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Mon, 25 Oct 2021 20:18:10 -0400
Subject: Oct 26 at 12pm (Zoom) -- Chen Dan (CMU CSD) -- Sharp Statistical
 Guarantees for Adversarially Robust Gaussian Classification -- AI
 Seminar sponsored by Morgan Stanley
In-Reply-To: <CAJA0PPAeb_9y28LwVKY75NbO3-D0SJnu2U94uk-BxuY8_t3dyw@mail.gmail.com>
References: <CAJA0PPAeb_9y28LwVKY75NbO3-D0SJnu2U94uk-BxuY8_t3dyw@mail.gmail.com>
Message-ID: <CAJA0PPAntGVxsA3YquCiV1m7nrHsjUtaHcvr1Obn64f3TbXzeQ@mail.gmail.com>

Hi all,

Just a reminder that the CMU AI Seminar is tomorrow 12pm-1pm:
https://cmu.zoom.us/j/93155268338?pwd=VVZYTFFEMTNLZlJVY1NmU1c3cXUzZz09   .

Chen Dan (CMU CSD) will be talking about his recent work on statistical
understanding of adversarial robustness.

Thanks,
Shaojie

On Sat, Oct 23, 2021 at 3:08 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (10/26)* from *1**2:00-1:00
> PM (U.S. Eastern time)* for the next talk of our *CMU AI seminar*,
> sponsored by Morgan Stanley
> <https://www.morganstanley.com/about-us/technology/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>
> On 10/26, *Chen Dan* (CMU CSD) will be giving a talk on "*Sharp
> Statistical Guarantees for Adversarially Robust Gaussian Classification*".
>
> *Title*: Sharp Statistical Guarantees for Adversarially Robust Gaussian
> Classification
>
> *Talk Abstract*: Adversarial robustness has become a fundamental
> requirement in modern machine learning applications. Yet, there has been
> surprisingly little statistical understanding so far. In this work, we
> provide the first result of the optimal minimax guarantees for the excess
> risk for adversarially robust classification, under a Gaussian mixture
> model studied by Schmidt et al. 2018. The results are stated in terms of
> the Adversarial Signal-to-Noise Ratio (AdvSNR), which generalizes a similar
> notion for standard linear classification to the adversarial setting. We
> establish an excess risk lower bound and design a computationally efficient
> estimator that achieves this optimal rate. Our results built upon a minimal
> set of assumptions while covering a wide spectrum of adversarial
> perturbations including L_p balls for any p>1. Joint work with Yuting Wei
> and Pradeep Ravikumar.
>
> *Speaker Bio*:  Chen Dan is a 6th year Ph.D. student at Computer Science
> Department, Carnegie Mellon University, advised by Pradeep Ravikumar. His
> research interest is in the broad area of robust statistical learning, with
> an emphasis on the theoretical understanding and practical algorithms for
> learning under various types of adversarial distribution shift. Prior to
> joining CMU, Chen received his bachelor degree from School of EECS, Peking
> University in 2016.
>
> *Zoom Link*:
> https://cmu.zoom.us/j/93155268338?pwd=VVZYTFFEMTNLZlJVY1NmU1c3cXUzZz09
>
> Thanks,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211025/c0d69c77/attachment.html>

From shaojieb at cs.cmu.edu  Tue Oct 26 12:03:25 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Tue, 26 Oct 2021 12:03:25 -0400
Subject: Oct 26 at 12pm (Zoom) -- Chen Dan (CMU CSD) -- Sharp Statistical
 Guarantees for Adversarially Robust Gaussian Classification -- AI
 Seminar sponsored by Morgan Stanley
In-Reply-To: <CAJA0PPAntGVxsA3YquCiV1m7nrHsjUtaHcvr1Obn64f3TbXzeQ@mail.gmail.com>
References: <CAJA0PPAeb_9y28LwVKY75NbO3-D0SJnu2U94uk-BxuY8_t3dyw@mail.gmail.com>
 <CAJA0PPAntGVxsA3YquCiV1m7nrHsjUtaHcvr1Obn64f3TbXzeQ@mail.gmail.com>
Message-ID: <CAJA0PPDOd5Cz9-DTb6s0nVVt3NepGc85UMY7m7Wwz1CTrp=6nw@mail.gmail.com>

Hi all,

The seminar today by Chen Dan on adversarial robustness is happening right
now!

In case you are interested:
https://cmu.zoom.us/j/93155268338?pwd=VVZYTFFEMTNLZlJVY1NmU1c3cXUzZz09

Best,
Shaojie

On Mon, Oct 25, 2021 at 8:18 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:

> Hi all,
>
> Just a reminder that the CMU AI Seminar is tomorrow 12pm-1pm:
> https://cmu.zoom.us/j/93155268338?pwd=VVZYTFFEMTNLZlJVY1NmU1c3cXUzZz09   .
>
> Chen Dan (CMU CSD) will be talking about his recent work on statistical
> understanding of adversarial robustness.
>
> Thanks,
> Shaojie
>
> On Sat, Oct 23, 2021 at 3:08 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:
>
>> Dear all,
>>
>> We look forward to seeing you *next Tuesday (10/26)* from *1**2:00-1:00
>> PM (U.S. Eastern time)* for the next talk of our *CMU AI seminar*,
>> sponsored by Morgan Stanley
>> <https://www.morganstanley.com/about-us/technology/>.
>>
>> To learn more about the seminar series or see the future schedule, please
>> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>>
>> On 10/26, *Chen Dan* (CMU CSD) will be giving a talk on "*Sharp
>> Statistical Guarantees for Adversarially Robust Gaussian Classification*
>> ".
>>
>> *Title*: Sharp Statistical Guarantees for Adversarially Robust Gaussian
>> Classification
>>
>> *Talk Abstract*: Adversarial robustness has become a fundamental
>> requirement in modern machine learning applications. Yet, there has been
>> surprisingly little statistical understanding so far. In this work, we
>> provide the first result of the optimal minimax guarantees for the excess
>> risk for adversarially robust classification, under a Gaussian mixture
>> model studied by Schmidt et al. 2018. The results are stated in terms of
>> the Adversarial Signal-to-Noise Ratio (AdvSNR), which generalizes a similar
>> notion for standard linear classification to the adversarial setting. We
>> establish an excess risk lower bound and design a computationally efficient
>> estimator that achieves this optimal rate. Our results built upon a minimal
>> set of assumptions while covering a wide spectrum of adversarial
>> perturbations including L_p balls for any p>1. Joint work with Yuting Wei
>> and Pradeep Ravikumar.
>>
>> *Speaker Bio*:  Chen Dan is a 6th year Ph.D. student at Computer Science
>> Department, Carnegie Mellon University, advised by Pradeep Ravikumar. His
>> research interest is in the broad area of robust statistical learning, with
>> an emphasis on the theoretical understanding and practical algorithms for
>> learning under various types of adversarial distribution shift. Prior to
>> joining CMU, Chen received his bachelor degree from School of EECS, Peking
>> University in 2016.
>>
>> *Zoom Link*:
>> https://cmu.zoom.us/j/93155268338?pwd=VVZYTFFEMTNLZlJVY1NmU1c3cXUzZz09
>>
>> Thanks,
>> Shaojie Bai (MLD)
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211026/1318ccbf/attachment.html>

From ashert at cs.cmu.edu  Wed Oct 27 10:56:12 2021
From: ashert at cs.cmu.edu (Asher Trockman)
Date: Wed, 27 Oct 2021 10:56:12 -0400
Subject: [CMU AI Seminar] Nov 2 at 12pm (Zoom) -- Jeremy Cohen (CMU MLD) --
 Gradient Descent on Neural Networks Typically Occurs at the Edge of Stability
 -- AI Seminar sponsored by Morgan Stanley
Message-ID: <CAMVmiv2oZWjyY0htXjWzC__==S1QijqEfW+36rHojaR1MzK=sw@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (11/2)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI Seminar*, sponsored
by Morgan Stanley <https://www.morganstanley.com/about-us/technology/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.

On 11/2, *Jeremy Cohen* (CMU MLD) will be giving a talk on "*Gradient
Descent on Neural Networks Typically Occurs at the Edge of Stability*".

*Title:* Gradient Descent on Neural Networks Typically Occurs at the Edge
of Stability

*Talk Abstract:* Neural networks are trained using optimization algorithms.
While we sometimes understand how these algorithms behave in restricted
settings (e.g. on quadratic or convex functions), very little is known
about the dynamics of these optimization algorithms on real neural
objective functions. In this paper, we take a close look at the simplest
optimization algorithm?full-batch gradient descent with a fixed step
size?and find that its behavior on neural networks is both (1) surprisingly
consistent across different architectures and tasks, and (2) surprisingly
different from that envisioned in the "conventional wisdom."

In particular, we empirically demonstrate that during gradient descent
training of neural networks, the maximum Hessian eigenvalue (the
"sharpness") always rises all the way to the largest stable value, which is
2/(step size), and then hovers just *above* that numerical value for the
remainder of training, in a regime we term the "Edge of Stability." (Click
here <https://twitter.com/deepcohen/status/1366881479175847942> for 1m 17s
animation.) At the Edge of Stability, the sharpness is still "trying" to
increase further?and that's what happens if you drop the step size?but is
somehow being actively restrained from doing so, by the implicit dynamics
of the optimization algorithm. Our findings have several implications for
the theory of neural network optimization. First, whereas the conventional
wisdom in optimization says that the sharpness ought to determine the step
size, our paper shows that in the topsy-turvy world of deep learning, the
reality is precisely the opposite: the *step size* wholly determines the
*sharpness*. Second, our findings imply that convergence analyses based on
L-smoothness, or on ensuring monotone descent, do not apply to neural
network training.

*Speaker Bio: *Jeremy Cohen is a PhD student in the Machine Learning
Department at CMU, co-advised by Zico Kolter and Ameet Talwalkar. His
research focus is "neural network plumbing": how to initialize and
normalize neural networks so that they train quickly and generalize well.

*Zoom Link:*
https://cmu.zoom.us/j/96099846691?pwd=NEc3UjQ4aHJ5dGhpTHpqYnQ2cnNaQT09

Thanks,
Asher Trockman
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211027/287427a5/attachment.html>

From ashert at cs.cmu.edu  Mon Nov  1 15:31:22 2021
From: ashert at cs.cmu.edu (Asher Trockman)
Date: Mon, 1 Nov 2021 15:31:22 -0400
Subject: [CMU AI Seminar] Nov 2 at 12pm (Zoom) -- Jeremy Cohen (CMU MLD)
 -- Gradient Descent on Neural Networks Typically Occurs at the Edge of
 Stability -- AI Seminar sponsored by Morgan Stanley
In-Reply-To: <CAMVmiv2oZWjyY0htXjWzC__==S1QijqEfW+36rHojaR1MzK=sw@mail.gmail.com>
References: <CAMVmiv2oZWjyY0htXjWzC__==S1QijqEfW+36rHojaR1MzK=sw@mail.gmail.com>
Message-ID: <CAMVmiv0NYDnKnNpCFd5rG2rH6269CsgfS+qYVcU=pniUFueLBA@mail.gmail.com>

Hi all,

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/>
is tomorrow *12pm-1pm*:
https://cmu.zoom.us/j/96099846691?pwd=NEc3UjQ4aHJ5dGhpTHpqYnQ2cnNaQT09.

*Jeremy Cohen (CMU MLD)* will be giving a talk on the surprising dynamics
of full-batch gradient descent on neural networks.

Thanks,
Asher


On Wed, Oct 27, 2021 at 10:56 AM Asher Trockman <ashert at cs.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (11/2)* from *1**2:00-1:00 PM
> (U.S. Eastern time)* for the next talk of our *CMU AI Seminar*, sponsored
> by Morgan Stanley <https://www.morganstanley.com/about-us/technology/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>
> On 11/2, *Jeremy Cohen* (CMU MLD) will be giving a talk on "*Gradient
> Descent on Neural Networks Typically Occurs at the Edge of Stability*".
>
> *Title:* Gradient Descent on Neural Networks Typically Occurs at the Edge
> of Stability
>
> *Talk Abstract:* Neural networks are trained using optimization
> algorithms. While we sometimes understand how these algorithms behave in
> restricted settings (e.g. on quadratic or convex functions), very little is
> known about the dynamics of these optimization algorithms on real neural
> objective functions. In this paper, we take a close look at the simplest
> optimization algorithm?full-batch gradient descent with a fixed step
> size?and find that its behavior on neural networks is both (1) surprisingly
> consistent across different architectures and tasks, and (2) surprisingly
> different from that envisioned in the "conventional wisdom."
>
> In particular, we empirically demonstrate that during gradient descent
> training of neural networks, the maximum Hessian eigenvalue (the
> "sharpness") always rises all the way to the largest stable value, which is
> 2/(step size), and then hovers just *above* that numerical value for the
> remainder of training, in a regime we term the "Edge of Stability." (Click
> here <https://twitter.com/deepcohen/status/1366881479175847942> for 1m
> 17s animation.) At the Edge of Stability, the sharpness is still "trying"
> to increase further?and that's what happens if you drop the step size?but
> is somehow being actively restrained from doing so, by the implicit
> dynamics of the optimization algorithm. Our findings have several
> implications for the theory of neural network optimization. First, whereas
> the conventional wisdom in optimization says that the sharpness ought to
> determine the step size, our paper shows that in the topsy-turvy world of
> deep learning, the reality is precisely the opposite: the *step size*
> wholly determines the *sharpness*. Second, our findings imply that
> convergence analyses based on L-smoothness, or on ensuring monotone
> descent, do not apply to neural network training.
>
> *Speaker Bio: *Jeremy Cohen is a PhD student in the Machine Learning
> Department at CMU, co-advised by Zico Kolter and Ameet Talwalkar. His
> research focus is "neural network plumbing": how to initialize and
> normalize neural networks so that they train quickly and generalize well.
>
> *Zoom Link:*
> https://cmu.zoom.us/j/96099846691?pwd=NEc3UjQ4aHJ5dGhpTHpqYnQ2cnNaQT09
>
> Thanks,
> Asher Trockman
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211101/93401f1c/attachment.html>

From ashert at cs.cmu.edu  Tue Nov  2 12:00:00 2021
From: ashert at cs.cmu.edu (Asher Trockman)
Date: Tue, 2 Nov 2021 12:00:00 -0400
Subject: [CMU AI Seminar] Nov 2 at 12pm (Zoom) -- Jeremy Cohen (CMU MLD)
 -- Gradient Descent on Neural Networks Typically Occurs at the Edge of
 Stability -- AI Seminar sponsored by Morgan Stanley
In-Reply-To: <CAMVmiv2oZWjyY0htXjWzC__==S1QijqEfW+36rHojaR1MzK=sw@mail.gmail.com>
References: <CAMVmiv2oZWjyY0htXjWzC__==S1QijqEfW+36rHojaR1MzK=sw@mail.gmail.com>
Message-ID: <CAMVmiv1vHRYLu648f0+TZ03i8d73Z-h_qiZCR7FDL8A7k4vkCg@mail.gmail.com>

Hi all,

The seminar today by Jeremy Cohen on "Gradient Descent on Neural Networks
Typically Occurs at the Edge of Stability" is happening now!

In case you are interested:
https://cmu.zoom.us/j/96099846691?pwd=NEc3UjQ4aHJ5dGhpTHpqYnQ2cnNaQT09

Best,
Asher

On Wed, Oct 27, 2021 at 10:56 AM Asher Trockman <ashert at cs.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (11/2)* from *1**2:00-1:00 PM
> (U.S. Eastern time)* for the next talk of our *CMU AI Seminar*, sponsored
> by Morgan Stanley <https://www.morganstanley.com/about-us/technology/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>
> On 11/2, *Jeremy Cohen* (CMU MLD) will be giving a talk on "*Gradient
> Descent on Neural Networks Typically Occurs at the Edge of Stability*".
>
> *Title:* Gradient Descent on Neural Networks Typically Occurs at the Edge
> of Stability
>
> *Talk Abstract:* Neural networks are trained using optimization
> algorithms. While we sometimes understand how these algorithms behave in
> restricted settings (e.g. on quadratic or convex functions), very little is
> known about the dynamics of these optimization algorithms on real neural
> objective functions. In this paper, we take a close look at the simplest
> optimization algorithm?full-batch gradient descent with a fixed step
> size?and find that its behavior on neural networks is both (1) surprisingly
> consistent across different architectures and tasks, and (2) surprisingly
> different from that envisioned in the "conventional wisdom."
>
> In particular, we empirically demonstrate that during gradient descent
> training of neural networks, the maximum Hessian eigenvalue (the
> "sharpness") always rises all the way to the largest stable value, which is
> 2/(step size), and then hovers just *above* that numerical value for the
> remainder of training, in a regime we term the "Edge of Stability." (Click
> here <https://twitter.com/deepcohen/status/1366881479175847942> for 1m
> 17s animation.) At the Edge of Stability, the sharpness is still "trying"
> to increase further?and that's what happens if you drop the step size?but
> is somehow being actively restrained from doing so, by the implicit
> dynamics of the optimization algorithm. Our findings have several
> implications for the theory of neural network optimization. First, whereas
> the conventional wisdom in optimization says that the sharpness ought to
> determine the step size, our paper shows that in the topsy-turvy world of
> deep learning, the reality is precisely the opposite: the *step size*
> wholly determines the *sharpness*. Second, our findings imply that
> convergence analyses based on L-smoothness, or on ensuring monotone
> descent, do not apply to neural network training.
>
> *Speaker Bio: *Jeremy Cohen is a PhD student in the Machine Learning
> Department at CMU, co-advised by Zico Kolter and Ameet Talwalkar. His
> research focus is "neural network plumbing": how to initialize and
> normalize neural networks so that they train quickly and generalize well.
>
> *Zoom Link:*
> https://cmu.zoom.us/j/96099846691?pwd=NEc3UjQ4aHJ5dGhpTHpqYnQ2cnNaQT09
>
> Thanks,
> Asher Trockman
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211102/8b0dda7f/attachment.html>

From shaojieb at cs.cmu.edu  Tue Nov  2 12:02:03 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Tue, 2 Nov 2021 12:02:03 -0400
Subject: [CMU AI Seminar] Nov 2 at 12pm (Zoom) -- Jeremy Cohen (CMU MLD)
 -- Gradient Descent on Neural Networks Typically Occurs at the Edge of
 Stability -- AI Seminar sponsored by Morgan Stanley
In-Reply-To: <CAMVmiv0NYDnKnNpCFd5rG2rH6269CsgfS+qYVcU=pniUFueLBA@mail.gmail.com>
References: <CAMVmiv2oZWjyY0htXjWzC__==S1QijqEfW+36rHojaR1MzK=sw@mail.gmail.com>
 <CAMVmiv0NYDnKnNpCFd5rG2rH6269CsgfS+qYVcU=pniUFueLBA@mail.gmail.com>
Message-ID: <CAJA0PPAVAjiM3twtQot3FH8TEqEjz9DoBgbiWHrZ6_etOUiAvg@mail.gmail.com>

Hi all,

*Jeremy Cohen*'s talk on surprising observations and dynamics of full-batch
GD on deep neural nets is starting in a few minutes!

Zoom link:
https://cmu.zoom.us/j/96099846691?pwd=NEc3UjQ4aHJ5dGhpTHpqYnQ2cnNaQT09

Best,
Shaojie

On Mon, Nov 1, 2021 at 3:32 PM Asher Trockman <ashert at cs.cmu.edu> wrote:

> Hi all,
>
> Just a reminder that the CMU AI Seminar
> <http://www.cs.cmu.edu/~aiseminar/> is tomorrow *12pm-1pm*:
> https://cmu.zoom.us/j/96099846691?pwd=NEc3UjQ4aHJ5dGhpTHpqYnQ2cnNaQT09.
>
> *Jeremy Cohen (CMU MLD)* will be giving a talk on the surprising dynamics
> of full-batch gradient descent on neural networks.
>
> Thanks,
> Asher
>
>
> On Wed, Oct 27, 2021 at 10:56 AM Asher Trockman <ashert at cs.cmu.edu> wrote:
>
>> Dear all,
>>
>> We look forward to seeing you *next Tuesday (11/2)* from *1**2:00-1:00
>> PM (U.S. Eastern time)* for the next talk of our *CMU AI Seminar*,
>> sponsored by Morgan Stanley
>> <https://www.morganstanley.com/about-us/technology/>.
>>
>> To learn more about the seminar series or see the future schedule, please
>> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>>
>> On 11/2, *Jeremy Cohen* (CMU MLD) will be giving a talk on "*Gradient
>> Descent on Neural Networks Typically Occurs at the Edge of Stability*".
>>
>> *Title:* Gradient Descent on Neural Networks Typically Occurs at the
>> Edge of Stability
>>
>> *Talk Abstract:* Neural networks are trained using optimization
>> algorithms. While we sometimes understand how these algorithms behave in
>> restricted settings (e.g. on quadratic or convex functions), very little is
>> known about the dynamics of these optimization algorithms on real neural
>> objective functions. In this paper, we take a close look at the simplest
>> optimization algorithm?full-batch gradient descent with a fixed step
>> size?and find that its behavior on neural networks is both (1) surprisingly
>> consistent across different architectures and tasks, and (2) surprisingly
>> different from that envisioned in the "conventional wisdom."
>>
>> In particular, we empirically demonstrate that during gradient descent
>> training of neural networks, the maximum Hessian eigenvalue (the
>> "sharpness") always rises all the way to the largest stable value, which is
>> 2/(step size), and then hovers just *above* that numerical value for the
>> remainder of training, in a regime we term the "Edge of Stability." (Click
>> here <https://twitter.com/deepcohen/status/1366881479175847942> for 1m
>> 17s animation.) At the Edge of Stability, the sharpness is still "trying"
>> to increase further?and that's what happens if you drop the step size?but
>> is somehow being actively restrained from doing so, by the implicit
>> dynamics of the optimization algorithm. Our findings have several
>> implications for the theory of neural network optimization. First, whereas
>> the conventional wisdom in optimization says that the sharpness ought to
>> determine the step size, our paper shows that in the topsy-turvy world of
>> deep learning, the reality is precisely the opposite: the *step size*
>> wholly determines the *sharpness*. Second, our findings imply that
>> convergence analyses based on L-smoothness, or on ensuring monotone
>> descent, do not apply to neural network training.
>>
>> *Speaker Bio: *Jeremy Cohen is a PhD student in the Machine Learning
>> Department at CMU, co-advised by Zico Kolter and Ameet Talwalkar. His
>> research focus is "neural network plumbing": how to initialize and
>> normalize neural networks so that they train quickly and generalize well.
>>
>> *Zoom Link:*
>> https://cmu.zoom.us/j/96099846691?pwd=NEc3UjQ4aHJ5dGhpTHpqYnQ2cnNaQT09
>>
>> Thanks,
>> Asher Trockman
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211102/8544878c/attachment-0001.html>

From shaojieb at cs.cmu.edu  Fri Nov  5 12:07:13 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Fri, 5 Nov 2021 12:07:13 -0400
Subject: [CMU AI Seminar] Nov 9 at 12pm (Zoom) -- Quanquan Gu (UCLA) --
 Stochastic Gradient Descent: Benign Overfitting and Implicit Regularization
 -- AI Seminar sponsored by Morgan Stanley
Message-ID: <CAJA0PPDfTJUdjjAD0oauS2ODtfyLTYU+emb9vSWN5t_UftufOQ@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (11/9)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI Seminar*, sponsored
by Morgan Stanley <https://www.morganstanley.com/about-us/technology/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.

On 11/9, *Quanquan Gu* (UCLA) will be giving a talk on "*Stochastic
Gradient Descent: Benign Overfitting and Implicit Regularization*" and his
group's latest research progress on DL theory.

*Title:* Stochastic Gradient Descent: Benign Overfitting and Implicit
Regularization

*Talk Abstract:* There is an increasing realization that algorithmic
inductive biases are central in preventing overfitting; empirically, we
often see a benign overfitting phenomenon in overparameterized settings for
natural learning algorithms, such as stochastic gradient descent (SGD),
where little to no explicit regularization has been employed. In the first
part of this talk, I will discuss benign overfitting of constant-stepsize
SGD in arguably the most basic setting: linear regression in the
overparameterized regime. Our main results provide a sharp excess risk
bound, stated in terms of the full eigenspectrum of the data covariance
matrix, that reveals a bias-variance decomposition characterizing when
generalization is possible. In the second part of this talk, I will
introduce sharp instance-based comparisons of the implicit regularization
of SGD with the explicit regularization of ridge regression, which are
conducted in a sample-inflation manner. I will show that provided up to
polylogarithmically more sample size, the generalization performance of SGD
is always no worse than that of ridge regression for a broad class of least
squares problem instances, and could be much better for some problem
instances. This suggests the benefits of implicit regularization in SGD
compared with the explicit regularization of ridge regression. This is
joint work with Difan Zou, Jingfeng Wu, Vladimir Braverman, Dean P. Foster
and Sham M. Kakade.

*Speaker Bio: *Quanquan Gu is an Assistant Professor of Computer Science at
UCLA. His research is in the area of artificial intelligence and machine
learning, with a focus on developing and analyzing nonconvex optimization
algorithms for machine learning to understand large-scale, dynamic,
complex, and heterogeneous data and building the theoretical foundations of
deep learning and reinforcement learning. He received his Ph.D. degree in
Computer Science from the University of Illinois at Urbana-Champaign in
2014. He is a recipient of the NSF CAREER Award, Simons Berkeley Research
Fellowship among other industrial research awards.

*Zoom Link: *
https://cmu.zoom.us/j/97788824898?pwd=alM4T1EvK1VHdEZ6aWdOa0lWOHdrZz09
<https://www.google.com/url?q=https://cmu.zoom.us/j/97788824898?pwd%3DalM4T1EvK1VHdEZ6aWdOa0lWOHdrZz09&sa=D&source=calendar&ust=1636461071062736&usg=AOvVaw1uuw9SRd6OXo_LaYjjKNvz>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211105/b1b689ba/attachment.html>

From shaojieb at cs.cmu.edu  Mon Nov  8 14:43:56 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Mon, 8 Nov 2021 14:43:56 -0500
Subject: [CMU AI Seminar] Nov 9 at 12pm (Zoom) -- Quanquan Gu (UCLA) --
 Stochastic Gradient Descent: Benign Overfitting and Implicit Regularization
 -- AI Seminar sponsored by Morgan Stanley
In-Reply-To: <CAJA0PPDfTJUdjjAD0oauS2ODtfyLTYU+emb9vSWN5t_UftufOQ@mail.gmail.com>
References: <CAJA0PPDfTJUdjjAD0oauS2ODtfyLTYU+emb9vSWN5t_UftufOQ@mail.gmail.com>
Message-ID: <CAJA0PPCi8nbjLfwxjQU0=Mu_hu=JO+ZYSaa7DfPWU6YD+MdQUA@mail.gmail.com>

Dear all,

Just a reminder that the CMU AI Seminar is tomorrow 12pm-1pm:
https://cmu.zoom.us/j/97788824898?pwd=alM4T1EvK1VHdEZ6aWdOa0lWOHdrZz09
<https://www.google.com/url?q=https://cmu.zoom.us/j/97788824898?pwd%3DalM4T1EvK1VHdEZ6aWdOa0lWOHdrZz09&sa=D&source=calendar&ust=1636461071062736&usg=AOvVaw1uuw9SRd6OXo_LaYjjKNvz>
.

*Professor Quanquan Gu (UCLA) *will be giving a talk on some
surprising findings, such as the implicit regularization effect, of SGD.

Thanks,
Asher

On Fri, Nov 5, 2021 at 12:07 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (11/9)* from *1**2:00-1:00 PM
> (U.S. Eastern time)* for the next talk of our *CMU AI Seminar*, sponsored
> by Morgan Stanley <https://www.morganstanley.com/about-us/technology/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>
> On 11/9, *Quanquan Gu* (UCLA) will be giving a talk on "*Stochastic
> Gradient Descent: Benign Overfitting and Implicit Regularization*" and
> his group's latest research progress on DL theory.
>
> *Title:* Stochastic Gradient Descent: Benign Overfitting and Implicit
> Regularization
>
> *Talk Abstract:* There is an increasing realization that algorithmic
> inductive biases are central in preventing overfitting; empirically, we
> often see a benign overfitting phenomenon in overparameterized settings for
> natural learning algorithms, such as stochastic gradient descent (SGD),
> where little to no explicit regularization has been employed. In the first
> part of this talk, I will discuss benign overfitting of constant-stepsize
> SGD in arguably the most basic setting: linear regression in the
> overparameterized regime. Our main results provide a sharp excess risk
> bound, stated in terms of the full eigenspectrum of the data covariance
> matrix, that reveals a bias-variance decomposition characterizing when
> generalization is possible. In the second part of this talk, I will
> introduce sharp instance-based comparisons of the implicit regularization
> of SGD with the explicit regularization of ridge regression, which are
> conducted in a sample-inflation manner. I will show that provided up to
> polylogarithmically more sample size, the generalization performance of SGD
> is always no worse than that of ridge regression for a broad class of least
> squares problem instances, and could be much better for some problem
> instances. This suggests the benefits of implicit regularization in SGD
> compared with the explicit regularization of ridge regression. This is
> joint work with Difan Zou, Jingfeng Wu, Vladimir Braverman, Dean P. Foster
> and Sham M. Kakade.
>
> *Speaker Bio: *Quanquan Gu is an Assistant Professor of Computer Science
> at UCLA. His research is in the area of artificial intelligence and machine
> learning, with a focus on developing and analyzing nonconvex optimization
> algorithms for machine learning to understand large-scale, dynamic,
> complex, and heterogeneous data and building the theoretical foundations of
> deep learning and reinforcement learning. He received his Ph.D. degree in
> Computer Science from the University of Illinois at Urbana-Champaign in
> 2014. He is a recipient of the NSF CAREER Award, Simons Berkeley Research
> Fellowship among other industrial research awards.
>
> *Zoom Link: *
> https://cmu.zoom.us/j/97788824898?pwd=alM4T1EvK1VHdEZ6aWdOa0lWOHdrZz09
> <https://www.google.com/url?q=https://cmu.zoom.us/j/97788824898?pwd%3DalM4T1EvK1VHdEZ6aWdOa0lWOHdrZz09&sa=D&source=calendar&ust=1636461071062736&usg=AOvVaw1uuw9SRd6OXo_LaYjjKNvz>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211108/09ea6007/attachment.html>

From shaojieb at cs.cmu.edu  Tue Nov  9 12:02:18 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Tue, 9 Nov 2021 12:02:18 -0500
Subject: [CMU AI Seminar] Nov 9 at 12pm (Zoom) -- Quanquan Gu (UCLA) --
 Stochastic Gradient Descent: Benign Overfitting and Implicit Regularization
 -- AI Seminar sponsored by Morgan Stanley
In-Reply-To: <CAJA0PPCi8nbjLfwxjQU0=Mu_hu=JO+ZYSaa7DfPWU6YD+MdQUA@mail.gmail.com>
References: <CAJA0PPDfTJUdjjAD0oauS2ODtfyLTYU+emb9vSWN5t_UftufOQ@mail.gmail.com>
 <CAJA0PPCi8nbjLfwxjQU0=Mu_hu=JO+ZYSaa7DfPWU6YD+MdQUA@mail.gmail.com>
Message-ID: <CAJA0PPAo3h=oMTJ2NBR0rdw7Z-OVKTKGpahi-ZGFJKEdgq6Riw@mail.gmail.com>

Hi all,

Quanquan Gu's (UCLA) talk on the surprising effect of SGD is starting right
now!

Best,
Shaojie

On Mon, Nov 8, 2021 at 2:43 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:

> Dear all,
>
> Just a reminder that the CMU AI Seminar is tomorrow 12pm-1pm:
> https://cmu.zoom.us/j/97788824898?pwd=alM4T1EvK1VHdEZ6aWdOa0lWOHdrZz09
> <https://www.google.com/url?q=https://cmu.zoom.us/j/97788824898?pwd%3DalM4T1EvK1VHdEZ6aWdOa0lWOHdrZz09&sa=D&source=calendar&ust=1636461071062736&usg=AOvVaw1uuw9SRd6OXo_LaYjjKNvz>
> .
>
> *Professor Quanquan Gu (UCLA) *will be giving a talk on some
> surprising findings, such as the implicit regularization effect, of SGD.
>
> Thanks,
> Asher
>
> On Fri, Nov 5, 2021 at 12:07 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:
>
>> Dear all,
>>
>> We look forward to seeing you *next Tuesday (11/9)* from *1**2:00-1:00
>> PM (U.S. Eastern time)* for the next talk of our *CMU AI Seminar*,
>> sponsored by Morgan Stanley
>> <https://www.morganstanley.com/about-us/technology/>.
>>
>> To learn more about the seminar series or see the future schedule,
>> please visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>>
>> On 11/9, *Quanquan Gu* (UCLA) will be giving a talk on "*Stochastic
>> Gradient Descent: Benign Overfitting and Implicit Regularization*" and
>> his group's latest research progress on DL theory.
>>
>> *Title:* Stochastic Gradient Descent: Benign Overfitting and Implicit
>> Regularization
>>
>> *Talk Abstract:* There is an increasing realization that algorithmic
>> inductive biases are central in preventing overfitting; empirically, we
>> often see a benign overfitting phenomenon in overparameterized settings for
>> natural learning algorithms, such as stochastic gradient descent (SGD),
>> where little to no explicit regularization has been employed. In the first
>> part of this talk, I will discuss benign overfitting of constant-stepsize
>> SGD in arguably the most basic setting: linear regression in the
>> overparameterized regime. Our main results provide a sharp excess risk
>> bound, stated in terms of the full eigenspectrum of the data covariance
>> matrix, that reveals a bias-variance decomposition characterizing when
>> generalization is possible. In the second part of this talk, I will
>> introduce sharp instance-based comparisons of the implicit regularization
>> of SGD with the explicit regularization of ridge regression, which are
>> conducted in a sample-inflation manner. I will show that provided up to
>> polylogarithmically more sample size, the generalization performance of SGD
>> is always no worse than that of ridge regression for a broad class of least
>> squares problem instances, and could be much better for some problem
>> instances. This suggests the benefits of implicit regularization in SGD
>> compared with the explicit regularization of ridge regression. This is
>> joint work with Difan Zou, Jingfeng Wu, Vladimir Braverman, Dean P. Foster
>> and Sham M. Kakade.
>>
>> *Speaker Bio: *Quanquan Gu is an Assistant Professor of Computer Science
>> at UCLA. His research is in the area of artificial intelligence and machine
>> learning, with a focus on developing and analyzing nonconvex optimization
>> algorithms for machine learning to understand large-scale, dynamic,
>> complex, and heterogeneous data and building the theoretical foundations of
>> deep learning and reinforcement learning. He received his Ph.D. degree in
>> Computer Science from the University of Illinois at Urbana-Champaign in
>> 2014. He is a recipient of the NSF CAREER Award, Simons Berkeley Research
>> Fellowship among other industrial research awards.
>>
>> *Zoom Link: *
>> https://cmu.zoom.us/j/97788824898?pwd=alM4T1EvK1VHdEZ6aWdOa0lWOHdrZz09
>> <https://www.google.com/url?q=https://cmu.zoom.us/j/97788824898?pwd%3DalM4T1EvK1VHdEZ6aWdOa0lWOHdrZz09&sa=D&source=calendar&ust=1636461071062736&usg=AOvVaw1uuw9SRd6OXo_LaYjjKNvz>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211109/c7847f64/attachment.html>

From shaojieb at cs.cmu.edu  Fri Nov 12 11:29:03 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Fri, 12 Nov 2021 11:29:03 -0500
Subject: [CMU AI Seminar] Nov 16 at 12pm (Zoom) -- Moritz Hardt (UC Berkeley)
 -- Retiring Adult: New Datasets for Fair Machine Learning -- AI
 Seminar sponsored by Morgan Stanley
Message-ID: <CAJA0PPBzArTL4rhi+R15+0mMgtYoQebSrjD101d9QS1gn8f3aQ@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (11/16)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI Seminar*, sponsored
by Morgan Stanley <https://www.morganstanley.com/about-us/technology/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.

On 11/16, *Moritz Hardt* (UC Berkeley) will be giving a talk on "*Retiring
Adult: New Datasets for Fair Machine Learning*" and his latest research on
fair machine learning.

*Title:* Retiring Adult: New Datasets for Fair Machine Learning

*Talk Abstract:* Although the fairness community has recognized the
importance of data, researchers in the area primarily rely on UCI Adult
when it comes to tabular data. Derived from a 1994 US Census survey, this
dataset has appeared in hundreds of research papers where it served as the
basis for the development and comparison of many algorithmic fairness
interventions. We reconstruct a superset of the UCI Adult data from
available US Census sources and reveal idiosyncrasies of the UCI Adult
dataset that limit its external validity. Our primary contribution is a
suite of new datasets derived from US Census surveys that extend the
existing data ecosystem for research on fair machine learning. We create
prediction tasks relating to income, employment, health, transportation,
and housing. The data span multiple years and all states of the United
States, allowing researchers to study temporal shift and geographic
variation. We highlight a broad initial sweep of new empirical insights
relating to trade-offs between fairness criteria, performance of
algorithmic interventions, and the role of distribution shift based on our
new datasets. Our findings inform ongoing debates, challenge some existing
narratives, and point to future research directions. Our datasets are
available at folktables.org.

*Speaker Bio: *Moritz Hardt is an Assistant Professor in the Department of
Electrical Engineering and Computer Sciences at the University of
California, Berkeley. His work builds foundations of machine learning and
algorithmic decision making with a focus on social context, interaction,
and impact. Hardt obtained a PhD in Computer Science from Princeton
University with a dissertation on privacy-preserving data analysis and
fairness in classification. He then held research positions at IBM Research
and Google. Hardt co-founded the Workshop on Fairness, Accountability, and
Transparency in Machine Learning. He is a co-author of "Fairness and
Machine Learning: Limitations and Opportunities" and "Patterns,
Predictions, and Actions: A Story about Machine Learning". He has received
an NSF CAREER award, a Sloan fellowship, and best paper awards at ICML 2018
and ICLR 2017.

*Zoom Link: *
https://cmu.zoom.us/j/99599979949?pwd=YXdzek1ic1FTbkZ6RytaN09Vajdodz09

Best,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211112/804ce2d4/attachment.html>

From shaojieb at cs.cmu.edu  Mon Nov 15 16:12:29 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Mon, 15 Nov 2021 16:12:29 -0500
Subject: [CMU AI Seminar] Nov 16 at 12pm (Zoom) -- Moritz Hardt (UC
 Berkeley) -- Retiring Adult: New Datasets for Fair Machine Learning -- AI
 Seminar sponsored by Morgan Stanley
In-Reply-To: <CAJA0PPBzArTL4rhi+R15+0mMgtYoQebSrjD101d9QS1gn8f3aQ@mail.gmail.com>
References: <CAJA0PPBzArTL4rhi+R15+0mMgtYoQebSrjD101d9QS1gn8f3aQ@mail.gmail.com>
Message-ID: <CAJA0PPAyyO_dSwzhtAReN-PYUoXJNaJoHq3OVRVkPWpRptnfSQ@mail.gmail.com>

Hi all,

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow *12pm-1pm*:
https://cmu.zoom.us/j/99599979949?pwd=YXdzek1ic1FTbkZ6RytaN09Vajdodz09.

*Professor Moritz Hardt (UC Berkeley) *will be giving a talk on the
fairness in ML and a new dataset they recently worked on.

(See (updated) talk abstract below)

Thanks,
Shaojie

-----------------------------------------
*Talk Abstract:* Although the fairness community has recognized the
importance of data, researchers in the area primarily rely on UCI Adult
when it comes to tabular data. Derived from a 1994 US Census survey, this
dataset has appeared in hundreds of research papers where it served as the
basis for the development and comparison of many algorithmic fairness
interventions. We reconstruct a superset of the UCI Adult data from
available US Census sources and reveal idiosyncrasies of the UCI Adult
dataset that limit its external validity. Our primary contribution is a
suite of new datasets derived from US Census surveys that extend the
existing data ecosystem for research on fair machine learning. We create
prediction tasks relating to income, employment, health, transportation,
and housing. The data span multiple years and all states of the United
States, allowing researchers to study temporal shift and geographic
variation. We highlight a broad initial sweep of new empirical insights
relating to trade-offs between fairness criteria, performance of
algorithmic interventions, and the role of distribution shift based on our
new datasets. Our findings inform ongoing debates, challenge some existing
narratives, and point to future research directions. Our datasets are
available at folktables.org. Joint work with Frances Ding, John Miller, and
Ludwig Schmidt.

On Fri, Nov 12, 2021 at 11:29 AM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (11/16)* from *1**2:00-1:00
> PM (U.S. Eastern time)* for the next talk of our *CMU AI Seminar*,
> sponsored by Morgan Stanley
> <https://www.morganstanley.com/about-us/technology/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>
> On 11/16, *Moritz Hardt* (UC Berkeley) will be giving a talk on "*Retiring
> Adult: New Datasets for Fair Machine Learning*" and his latest research
> on fair machine learning.
>
> *Title:* Retiring Adult: New Datasets for Fair Machine Learning
>
> *Talk Abstract:* Although the fairness community has recognized the
> importance of data, researchers in the area primarily rely on UCI Adult
> when it comes to tabular data. Derived from a 1994 US Census survey, this
> dataset has appeared in hundreds of research papers where it served as the
> basis for the development and comparison of many algorithmic fairness
> interventions. We reconstruct a superset of the UCI Adult data from
> available US Census sources and reveal idiosyncrasies of the UCI Adult
> dataset that limit its external validity. Our primary contribution is a
> suite of new datasets derived from US Census surveys that extend the
> existing data ecosystem for research on fair machine learning. We create
> prediction tasks relating to income, employment, health, transportation,
> and housing. The data span multiple years and all states of the United
> States, allowing researchers to study temporal shift and geographic
> variation. We highlight a broad initial sweep of new empirical insights
> relating to trade-offs between fairness criteria, performance of
> algorithmic interventions, and the role of distribution shift based on our
> new datasets. Our findings inform ongoing debates, challenge some existing
> narratives, and point to future research directions. Our datasets are
> available at folktables.org.
>
> *Speaker Bio: *Moritz Hardt is an Assistant Professor in the Department
> of Electrical Engineering and Computer Sciences at the University of
> California, Berkeley. His work builds foundations of machine learning and
> algorithmic decision making with a focus on social context, interaction,
> and impact. Hardt obtained a PhD in Computer Science from Princeton
> University with a dissertation on privacy-preserving data analysis and
> fairness in classification. He then held research positions at IBM Research
> and Google. Hardt co-founded the Workshop on Fairness, Accountability, and
> Transparency in Machine Learning. He is a co-author of "Fairness and
> Machine Learning: Limitations and Opportunities" and "Patterns,
> Predictions, and Actions: A Story about Machine Learning". He has received
> an NSF CAREER award, a Sloan fellowship, and best paper awards at ICML 2018
> and ICLR 2017.
>
> *Zoom Link: *
> https://cmu.zoom.us/j/99599979949?pwd=YXdzek1ic1FTbkZ6RytaN09Vajdodz09
>
> Best,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211115/6a197970/attachment.html>

From shaojieb at cs.cmu.edu  Fri Nov 19 12:51:46 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Fri, 19 Nov 2021 12:51:46 -0500
Subject: [CMU AI Seminar] Nov 23 at 12pm (Zoom) -- Eric Wallace (UC Berkeley)
 -- What Can We Learn From Vulnerabilities of ML Models? -- AI Seminar
 sponsored by Morgan Stanley
Message-ID: <CAJA0PPB5FA_a2bnCfg1rkzunKW8jpEM+zTMRSzT4xoeoY6=GKg@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (11/23)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI Seminar*, sponsored
by Morgan Stanley <https://www.morganstanley.com/about-us/technology/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.

On 11/23, *Eric Wallace* (UC Berkeley) will be giving a talk on "*What can
we learn from vulnerabilities of ML models*" and sharing his latest
research on large-scale NLP models and their vulnerabilities.

*Title:*  What Can We Learn From Vulnerabilities of ML Models?

*Talk Abstract:* Today's neural network models achieve high accuracy on
in-distribution data and are being widely deployed in production systems.
This talk will discuss attacks on such models that not only expose
worrisome security and privacy vulnerabilities, but also provide new
perspectives into how and why the models work. Concretely, I will show how
realistic adversaries can extract secret training data, steal model
weights, and manipulate test predictions, all using black-box access to
models at either training- or test-time. These attacks will reveal
different insights, including how NLP models rely on dataset biases and
spurious correlations, and how their training dynamics impact memorization
of examples. Finally, I will discuss defenses against these vulnerabilities
and suggest practical takeaways for developing secure ML systems.

*Speaker Bio: *Eric Wallace is a 3rd year PhD student at UC Berkeley
advised by Dan Klein and Dawn Song. His research interests center around
large language models and making them more secure, private, and robust.
Eric's work received the best demo award at EMNLP 2019.

*Zoom Link: *
https://cmu.zoom.us/j/96673560117?pwd=WHFMWERjWkphbnlNMWl2cmk5aE1QZz09

Best,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211119/e459649e/attachment.html>

From shaojieb at cs.cmu.edu  Mon Nov 22 16:10:57 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Mon, 22 Nov 2021 16:10:57 -0500
Subject: [CMU AI Seminar] Nov 23 at 12pm (Zoom) -- Eric Wallace (UC
 Berkeley) -- What Can We Learn From Vulnerabilities of ML Models? -- AI
 Seminar sponsored by Morgan Stanley
In-Reply-To: <CAJA0PPB5FA_a2bnCfg1rkzunKW8jpEM+zTMRSzT4xoeoY6=GKg@mail.gmail.com>
References: <CAJA0PPB5FA_a2bnCfg1rkzunKW8jpEM+zTMRSzT4xoeoY6=GKg@mail.gmail.com>
Message-ID: <CAJA0PPB7FVTFD+k54i3iSJchLC_RTzAhPf1yi_LqpYvqkKLpog@mail.gmail.com>

Hi all,

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow *12pm-1pm*:
https://cmu.zoom.us/j/96673560117?pwd=WHFMWERjWkphbnlNMWl2cmk5aE1QZz09.

*Eric Wallace* (UC Berkeley) will be talking about large-scale NLP models
and the their vulnerabilities (see below)!

--------------------------------------
*Title:*  What Can We Learn From Vulnerabilities of ML Models?

*Talk Abstract:* Today's neural network models achieve high accuracy on
in-distribution data and are being widely deployed in production systems.
This talk will discuss attacks on such models that not only expose
worrisome security and privacy vulnerabilities, but also provide new
perspectives into how and why the models work. Concretely, I will show how
realistic adversaries can extract secret training data, steal model
weights, and manipulate test predictions, all using black-box access to
models at either training- or test-time. These attacks will reveal
different insights, including how NLP models rely on dataset biases and
spurious correlations, and how their training dynamics impact memorization
of examples. Finally, I will discuss defenses against these vulnerabilities
and suggest practical takeaways for developing secure ML systems.

*Speaker Bio: *Eric Wallace is a 3rd year PhD student at UC Berkeley
advised by Dan Klein and Dawn Song. His research interests center around
large language models and making them more secure, private, and robust.
Eric's work received the best demo award at EMNLP 2019.
--------------------------------------

Thanks and Happy Thanksgiving,
Shaojie


On Fri, Nov 19, 2021 at 12:51 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (11/23)* from *1**2:00-1:00
> PM (U.S. Eastern time)* for the next talk of our *CMU AI Seminar*,
> sponsored by Morgan Stanley
> <https://www.morganstanley.com/about-us/technology/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>
> On 11/23, *Eric Wallace* (UC Berkeley) will be giving a talk on "*What
> can we learn from vulnerabilities of ML models*" and sharing his latest
> research on large-scale NLP models and their vulnerabilities.
>
> *Title:*  What Can We Learn From Vulnerabilities of ML Models?
>
> *Talk Abstract:* Today's neural network models achieve high accuracy on
> in-distribution data and are being widely deployed in production systems.
> This talk will discuss attacks on such models that not only expose
> worrisome security and privacy vulnerabilities, but also provide new
> perspectives into how and why the models work. Concretely, I will show how
> realistic adversaries can extract secret training data, steal model
> weights, and manipulate test predictions, all using black-box access to
> models at either training- or test-time. These attacks will reveal
> different insights, including how NLP models rely on dataset biases and
> spurious correlations, and how their training dynamics impact memorization
> of examples. Finally, I will discuss defenses against these vulnerabilities
> and suggest practical takeaways for developing secure ML systems.
>
> *Speaker Bio: *Eric Wallace is a 3rd year PhD student at UC Berkeley
> advised by Dan Klein and Dawn Song. His research interests center around
> large language models and making them more secure, private, and robust.
> Eric's work received the best demo award at EMNLP 2019.
>
> *Zoom Link: *
> https://cmu.zoom.us/j/96673560117?pwd=WHFMWERjWkphbnlNMWl2cmk5aE1QZz09
>
> Best,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211122/328d5d59/attachment.html>

From shaojieb at cs.cmu.edu  Tue Nov 23 12:01:29 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Tue, 23 Nov 2021 12:01:29 -0500
Subject: [CMU AI Seminar] Nov 23 at 12pm (Zoom) -- Eric Wallace (UC
 Berkeley) -- What Can We Learn From Vulnerabilities of ML Models? -- AI
 Seminar sponsored by Morgan Stanley
In-Reply-To: <CAJA0PPB7FVTFD+k54i3iSJchLC_RTzAhPf1yi_LqpYvqkKLpog@mail.gmail.com>
References: <CAJA0PPB5FA_a2bnCfg1rkzunKW8jpEM+zTMRSzT4xoeoY6=GKg@mail.gmail.com>
 <CAJA0PPB7FVTFD+k54i3iSJchLC_RTzAhPf1yi_LqpYvqkKLpog@mail.gmail.com>
Message-ID: <CAJA0PPC0N1vFHtQgtOO6R+dHTQV6u__9PtX9_jwqMGBBphCu0Q@mail.gmail.com>

Dear all,

Eric Wallace's (UC Berkeley) talk on large-scale NLP models and robustness
is starting now! Zoom:
https://cmu.zoom.us/j/96673560117?pwd=WHFMWERjWkphbnlNMWl2cmk5aE1QZz09

Best,
Shaojie

On Mon, Nov 22, 2021 at 4:10 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:

> Hi all,
>
> Just a reminder that the CMU AI Seminar
> <http://www.cs.cmu.edu/~aiseminar/> is tomorrow *12pm-1pm*:
> https://cmu.zoom.us/j/96673560117?pwd=WHFMWERjWkphbnlNMWl2cmk5aE1QZz09.
>
> *Eric Wallace* (UC Berkeley) will be talking about large-scale NLP models
> and the their vulnerabilities (see below)!
>
> --------------------------------------
> *Title:*  What Can We Learn From Vulnerabilities of ML Models?
>
> *Talk Abstract:* Today's neural network models achieve high accuracy on
> in-distribution data and are being widely deployed in production systems.
> This talk will discuss attacks on such models that not only expose
> worrisome security and privacy vulnerabilities, but also provide new
> perspectives into how and why the models work. Concretely, I will show how
> realistic adversaries can extract secret training data, steal model
> weights, and manipulate test predictions, all using black-box access to
> models at either training- or test-time. These attacks will reveal
> different insights, including how NLP models rely on dataset biases and
> spurious correlations, and how their training dynamics impact memorization
> of examples. Finally, I will discuss defenses against these vulnerabilities
> and suggest practical takeaways for developing secure ML systems.
>
> *Speaker Bio: *Eric Wallace is a 3rd year PhD student at UC Berkeley
> advised by Dan Klein and Dawn Song. His research interests center around
> large language models and making them more secure, private, and robust.
> Eric's work received the best demo award at EMNLP 2019.
> --------------------------------------
>
> Thanks and Happy Thanksgiving,
> Shaojie
>
>
> On Fri, Nov 19, 2021 at 12:51 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:
>
>> Dear all,
>>
>> We look forward to seeing you *next Tuesday (11/23)* from *1**2:00-1:00
>> PM (U.S. Eastern time)* for the next talk of our *CMU AI Seminar*,
>> sponsored by Morgan Stanley
>> <https://www.morganstanley.com/about-us/technology/>.
>>
>> To learn more about the seminar series or see the future schedule,
>> please visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>>
>> On 11/23, *Eric Wallace* (UC Berkeley) will be giving a talk on "*What
>> can we learn from vulnerabilities of ML models*" and sharing his latest
>> research on large-scale NLP models and their vulnerabilities.
>>
>> *Title:*  What Can We Learn From Vulnerabilities of ML Models?
>>
>> *Talk Abstract:* Today's neural network models achieve high accuracy on
>> in-distribution data and are being widely deployed in production systems.
>> This talk will discuss attacks on such models that not only expose
>> worrisome security and privacy vulnerabilities, but also provide new
>> perspectives into how and why the models work. Concretely, I will show how
>> realistic adversaries can extract secret training data, steal model
>> weights, and manipulate test predictions, all using black-box access to
>> models at either training- or test-time. These attacks will reveal
>> different insights, including how NLP models rely on dataset biases and
>> spurious correlations, and how their training dynamics impact memorization
>> of examples. Finally, I will discuss defenses against these vulnerabilities
>> and suggest practical takeaways for developing secure ML systems.
>>
>> *Speaker Bio: *Eric Wallace is a 3rd year PhD student at UC Berkeley
>> advised by Dan Klein and Dawn Song. His research interests center around
>> large language models and making them more secure, private, and robust.
>> Eric's work received the best demo award at EMNLP 2019.
>>
>> *Zoom Link: *
>> https://cmu.zoom.us/j/96673560117?pwd=WHFMWERjWkphbnlNMWl2cmk5aE1QZz09
>>
>> Best,
>> Shaojie Bai (MLD)
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211123/c04a4322/attachment.html>

From shaojieb at cs.cmu.edu  Fri Dec 10 13:46:13 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Fri, 10 Dec 2021 13:46:13 -0500
Subject: [CMU AI Seminar] Dec 14 at 12pm (Zoom) -- Christian Theobalt (Max
 Planck Institute for Informatics) -- Neural Methods for Reconstruction and
 Rendering of Real World Scenes -- AI Seminar sponsored by Morgan Stanley
Message-ID: <CAJA0PPCAKgCa3u4kmHQQ=nCLepMrEPFKjj_geccY4PJgRo9ewg@mail.gmail.com>

Dear all,

We look forward to seeing you *next Tuesday (12/14)* from *1**2:00-1:00 PM
(U.S. Eastern time)* for the next talk of our *CMU AI Seminar*, sponsored
by Morgan Stanley <https://www.morganstanley.com/about-us/technology/>.

To learn more about the seminar series or see the future schedule, please
visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.

On 12/14, Professor *Christian Theobalt* (Max Planck Institute for
Informatics) will be giving a talk on "*Neural Methods for Reconstruction
and Rendering of Real World Scenes*" and sharing his research on new neural
rendering approaches.

*Title:*  Neural Methods for Reconstruction and Rendering of Real World
Scenes

*Talk Abstract:* In this presentation, I will talk about some of the recent
work we did on new methods for reconstructing computer graphics models of
real world scenes from sparse or even monocular video data. These methods
are based of bringing together neural network-based and explicit
model-based approaches. I will also talk about new neural rendering
approaches that combine explicit model-based and neural network based
concepts for image formation in new ways. They enable new means to
synthesize highly realistic imagery and videos of real work scenes under
user control.

*Speaker Bio: *Christian Theobalt is The Scientific Director of the new
Visual Computing and Artificial Intelligence Department at the
Max-Planck-Institute for Informatics, Saarbr?cken, Germany. He is also a
Professor of Computer Science at Saarland University, Germany. From 2007
until 2009 he was a Visiting Assistant Professor in the Department of
Computer Science at Stanford University. He received his MSc degree in
Artificial Intelligence from the University of Edinburgh, his Diplom (MS)
degree in Computer Science from Saarland University, and his PhD (Dr.-Ing.)
from the Max-Planck-Institute for Informatics.


In his research he looks at algorithmic problems that lie at the
intersection of Computer Graphics, Computer Vision and Machine Learning,
such as: static and dynamic 3D scene reconstruction, neural rendering and
neural scene representations, marker-less motion and performance capture,
virtual humans, virtual and augmented reality, computer animation,
intrinsic video and inverse rendering, computational videography, machine
learning for graphics and vision, new sensors for 3D acquisition, as well
as image- and physically-based rendering. He is also interested in using
reconstruction techniques for human computer interaction. For his work, he
received several awards, including the Otto Hahn Medal of the Max-Planck
Society in 2007, the EUROGRAPHICS Young Researcher Award in 2009, the
German Pattern Recognition Award 2012, and the Karl Heinz Beckurts Award in
2017, and the EUROGRAPHICS Outstanding Technical Contributions Award in
2020. He received two ERC grants, an ERC Starting Grant in 2013 and an ERC
Consolidator Grant in 2017.

*Zoom Link: *
https://cmu.zoom.us/j/92289727241?pwd=a1VJUSsrYjJhblY1VEgzZFdMM2pLZz09

Best,
Shaojie Bai (MLD)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211210/aee63d6e/attachment.html>

From shaojieb at cs.cmu.edu  Mon Dec 13 12:28:25 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Mon, 13 Dec 2021 12:28:25 -0500
Subject: [CMU AI Seminar] Dec 14 at 12pm (Zoom) -- Christian Theobalt (Max
 Planck Institute for Informatics) -- Neural Methods for
 Reconstruction and
 Rendering of Real World Scenes -- AI Seminar sponsored by Morgan Stanley
In-Reply-To: <CAJA0PPCAKgCa3u4kmHQQ=nCLepMrEPFKjj_geccY4PJgRo9ewg@mail.gmail.com>
References: <CAJA0PPCAKgCa3u4kmHQQ=nCLepMrEPFKjj_geccY4PJgRo9ewg@mail.gmail.com>
Message-ID: <CAJA0PPDimt7DyV=7eJ1HODgGDErgYqb4+xA4A6Ujmz6y8Ty9Hw@mail.gmail.com>

Hi all,

Just a reminder that the CMU AI Seminar <http://www.cs.cmu.edu/~aiseminar/> is
tomorrow *12pm-1pm*:
https://cmu.zoom.us/j/92289727241?pwd=a1VJUSsrYjJhblY1VEgzZFdMM2pLZz09.

*Christian Theobalt* (Max Planck Institute for Informatics) will be talking
about neural rendering approaches!

--------------------------------------
*Title:*  Neural Methods for Reconstruction and Rendering of Real World
Scenes

*Talk Abstract:* In this presentation, I will talk about some of the recent
work we did on new methods for reconstructing computer graphics models of
real world scenes from sparse or even monocular video data. These methods
are based of bringing together neural network-based and explicit
model-based approaches. I will also talk about new neural rendering
approaches that combine explicit model-based and neural network based
concepts for image formation in new ways. They enable new means to
synthesize highly realistic imagery and videos of real work scenes under
user control.
--------------------------------------

Thanks,
Shaojie

On Fri, Dec 10, 2021 at 1:46 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (12/14)* from *1**2:00-1:00
> PM (U.S. Eastern time)* for the next talk of our *CMU AI Seminar*,
> sponsored by Morgan Stanley
> <https://www.morganstanley.com/about-us/technology/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>
> On 12/14, Professor *Christian Theobalt* (Max Planck Institute for
> Informatics) will be giving a talk on "*Neural Methods for Reconstruction
> and Rendering of Real World Scenes*" and sharing his research on new
> neural rendering approaches.
>
> *Title:*  Neural Methods for Reconstruction and Rendering of Real World
> Scenes
>
> *Talk Abstract:* In this presentation, I will talk about some of the
> recent work we did on new methods for reconstructing computer graphics
> models of real world scenes from sparse or even monocular video data. These
> methods are based of bringing together neural network-based and explicit
> model-based approaches. I will also talk about new neural rendering
> approaches that combine explicit model-based and neural network based
> concepts for image formation in new ways. They enable new means to
> synthesize highly realistic imagery and videos of real work scenes under
> user control.
>
> *Speaker Bio: *Christian Theobalt is The Scientific Director of the new
> Visual Computing and Artificial Intelligence Department at the
> Max-Planck-Institute for Informatics, Saarbr?cken, Germany. He is also a
> Professor of Computer Science at Saarland University, Germany. From 2007
> until 2009 he was a Visiting Assistant Professor in the Department of
> Computer Science at Stanford University. He received his MSc degree in
> Artificial Intelligence from the University of Edinburgh, his Diplom (MS)
> degree in Computer Science from Saarland University, and his PhD (Dr.-Ing.)
> from the Max-Planck-Institute for Informatics.
>
>
> In his research he looks at algorithmic problems that lie at the
> intersection of Computer Graphics, Computer Vision and Machine Learning,
> such as: static and dynamic 3D scene reconstruction, neural rendering and
> neural scene representations, marker-less motion and performance capture,
> virtual humans, virtual and augmented reality, computer animation,
> intrinsic video and inverse rendering, computational videography, machine
> learning for graphics and vision, new sensors for 3D acquisition, as well
> as image- and physically-based rendering. He is also interested in using
> reconstruction techniques for human computer interaction. For his work, he
> received several awards, including the Otto Hahn Medal of the Max-Planck
> Society in 2007, the EUROGRAPHICS Young Researcher Award in 2009, the
> German Pattern Recognition Award 2012, and the Karl Heinz Beckurts Award in
> 2017, and the EUROGRAPHICS Outstanding Technical Contributions Award in
> 2020. He received two ERC grants, an ERC Starting Grant in 2013 and an ERC
> Consolidator Grant in 2017.
>
> *Zoom Link: *
> https://cmu.zoom.us/j/92289727241?pwd=a1VJUSsrYjJhblY1VEgzZFdMM2pLZz09
>
> Best,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211213/22b0c42b/attachment.html>

From shaojieb at cs.cmu.edu  Mon Dec 13 16:10:22 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Mon, 13 Dec 2021 16:10:22 -0500
Subject: [CMU AI Seminar] Dec 14 at 12pm (Zoom) -- Christian Theobalt (Max
 Planck Institute for Informatics) -- Neural Methods for
 Reconstruction and
 Rendering of Real World Scenes -- AI Seminar sponsored by Morgan Stanley
In-Reply-To: <CAJA0PPDimt7DyV=7eJ1HODgGDErgYqb4+xA4A6Ujmz6y8Ty9Hw@mail.gmail.com>
References: <CAJA0PPCAKgCa3u4kmHQQ=nCLepMrEPFKjj_geccY4PJgRo9ewg@mail.gmail.com>
 <CAJA0PPDimt7DyV=7eJ1HODgGDErgYqb4+xA4A6Ujmz6y8Ty9Hw@mail.gmail.com>
Message-ID: <CAJA0PPAfr_vnBkHf_bOe9sYQgPukVU6NkW_SuvkswK5ptwsJdQ@mail.gmail.com>

Hi all,

Sorry for the spamming. But since numerous people have asked me about this
(as it's the end of the semester and people are traveling): *there will NOT
be a recording of the talk tomorrow* (upon the speaker's request).

Best,
Shaojie

On Mon, Dec 13, 2021 at 12:28 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:

> Hi all,
>
> Just a reminder that the CMU AI Seminar
> <http://www.cs.cmu.edu/~aiseminar/> is tomorrow *12pm-1pm*:
> https://cmu.zoom.us/j/92289727241?pwd=a1VJUSsrYjJhblY1VEgzZFdMM2pLZz09.
>
> *Christian Theobalt* (Max Planck Institute for Informatics) will be
> talking about neural rendering approaches!
>
> --------------------------------------
> *Title:*  Neural Methods for Reconstruction and Rendering of Real World
> Scenes
>
> *Talk Abstract:* In this presentation, I will talk about some of the
> recent work we did on new methods for reconstructing computer graphics
> models of real world scenes from sparse or even monocular video data. These
> methods are based of bringing together neural network-based and explicit
> model-based approaches. I will also talk about new neural rendering
> approaches that combine explicit model-based and neural network based
> concepts for image formation in new ways. They enable new means to
> synthesize highly realistic imagery and videos of real work scenes under
> user control.
> --------------------------------------
>
> Thanks,
> Shaojie
>
> On Fri, Dec 10, 2021 at 1:46 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:
>
>> Dear all,
>>
>> We look forward to seeing you *next Tuesday (12/14)* from *1**2:00-1:00
>> PM (U.S. Eastern time)* for the next talk of our *CMU AI Seminar*,
>> sponsored by Morgan Stanley
>> <https://www.morganstanley.com/about-us/technology/>.
>>
>> To learn more about the seminar series or see the future schedule,
>> please visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>>
>> On 12/14, Professor *Christian Theobalt* (Max Planck Institute for
>> Informatics) will be giving a talk on "*Neural Methods for
>> Reconstruction and Rendering of Real World Scenes*" and sharing his
>> research on new neural rendering approaches.
>>
>> *Title:*  Neural Methods for Reconstruction and Rendering of Real World
>> Scenes
>>
>> *Talk Abstract:* In this presentation, I will talk about some of the
>> recent work we did on new methods for reconstructing computer graphics
>> models of real world scenes from sparse or even monocular video data. These
>> methods are based of bringing together neural network-based and explicit
>> model-based approaches. I will also talk about new neural rendering
>> approaches that combine explicit model-based and neural network based
>> concepts for image formation in new ways. They enable new means to
>> synthesize highly realistic imagery and videos of real work scenes under
>> user control.
>>
>> *Speaker Bio: *Christian Theobalt is The Scientific Director of the new
>> Visual Computing and Artificial Intelligence Department at the
>> Max-Planck-Institute for Informatics, Saarbr?cken, Germany. He is also a
>> Professor of Computer Science at Saarland University, Germany. From 2007
>> until 2009 he was a Visiting Assistant Professor in the Department of
>> Computer Science at Stanford University. He received his MSc degree in
>> Artificial Intelligence from the University of Edinburgh, his Diplom (MS)
>> degree in Computer Science from Saarland University, and his PhD (Dr.-Ing.)
>> from the Max-Planck-Institute for Informatics.
>>
>>
>> In his research he looks at algorithmic problems that lie at the
>> intersection of Computer Graphics, Computer Vision and Machine Learning,
>> such as: static and dynamic 3D scene reconstruction, neural rendering and
>> neural scene representations, marker-less motion and performance capture,
>> virtual humans, virtual and augmented reality, computer animation,
>> intrinsic video and inverse rendering, computational videography, machine
>> learning for graphics and vision, new sensors for 3D acquisition, as well
>> as image- and physically-based rendering. He is also interested in using
>> reconstruction techniques for human computer interaction. For his work, he
>> received several awards, including the Otto Hahn Medal of the Max-Planck
>> Society in 2007, the EUROGRAPHICS Young Researcher Award in 2009, the
>> German Pattern Recognition Award 2012, and the Karl Heinz Beckurts Award in
>> 2017, and the EUROGRAPHICS Outstanding Technical Contributions Award in
>> 2020. He received two ERC grants, an ERC Starting Grant in 2013 and an ERC
>> Consolidator Grant in 2017.
>>
>> *Zoom Link: *
>> https://cmu.zoom.us/j/92289727241?pwd=a1VJUSsrYjJhblY1VEgzZFdMM2pLZz09
>>
>> Best,
>> Shaojie Bai (MLD)
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211213/fcd925b3/attachment-0001.html>

From shaojieb at cs.cmu.edu  Tue Dec 14 09:47:02 2021
From: shaojieb at cs.cmu.edu (Shaojie Bai)
Date: Tue, 14 Dec 2021 09:47:02 -0500
Subject: [CMU AI Seminar] Dec 14 at 12pm (Zoom) -- Christian Theobalt (Max
 Planck Institute for Informatics) -- Neural Methods for
 Reconstruction and
 Rendering of Real World Scenes -- AI Seminar sponsored by Morgan Stanley
In-Reply-To: <CAJA0PPCAKgCa3u4kmHQQ=nCLepMrEPFKjj_geccY4PJgRo9ewg@mail.gmail.com>
References: <CAJA0PPCAKgCa3u4kmHQQ=nCLepMrEPFKjj_geccY4PJgRo9ewg@mail.gmail.com>
Message-ID: <CAJA0PPDKvijAMLiyZeOGsQ3-VdCOvhNZYiT5PdEoA32vMtV9Fg@mail.gmail.com>

Hi all,

Sorry for the short notice, but our speaker, Christian, has a family
emergency and will not be able to attend the AI seminar as planned today at
12pm. *This event is thus CANCELLED*. He asked me to pass his apologies to
everyone who was planning to attend.

We will move Christian's talk to the spring semester.

Happy winter break, and see everyone in the spring!

Best,
Shaojie

On Fri, Dec 10, 2021 at 1:46 PM Shaojie Bai <shaojieb at cs.cmu.edu> wrote:

> Dear all,
>
> We look forward to seeing you *next Tuesday (12/14)* from *1**2:00-1:00
> PM (U.S. Eastern time)* for the next talk of our *CMU AI Seminar*,
> sponsored by Morgan Stanley
> <https://www.morganstanley.com/about-us/technology/>.
>
> To learn more about the seminar series or see the future schedule, please
> visit the seminar website <http://www.cs.cmu.edu/~aiseminar/>.
>
> On 12/14, Professor *Christian Theobalt* (Max Planck Institute for
> Informatics) will be giving a talk on "*Neural Methods for Reconstruction
> and Rendering of Real World Scenes*" and sharing his research on new
> neural rendering approaches.
>
> *Title:*  Neural Methods for Reconstruction and Rendering of Real World
> Scenes
>
> *Talk Abstract:* In this presentation, I will talk about some of the
> recent work we did on new methods for reconstructing computer graphics
> models of real world scenes from sparse or even monocular video data. These
> methods are based of bringing together neural network-based and explicit
> model-based approaches. I will also talk about new neural rendering
> approaches that combine explicit model-based and neural network based
> concepts for image formation in new ways. They enable new means to
> synthesize highly realistic imagery and videos of real work scenes under
> user control.
>
> *Speaker Bio: *Christian Theobalt is The Scientific Director of the new
> Visual Computing and Artificial Intelligence Department at the
> Max-Planck-Institute for Informatics, Saarbr?cken, Germany. He is also a
> Professor of Computer Science at Saarland University, Germany. From 2007
> until 2009 he was a Visiting Assistant Professor in the Department of
> Computer Science at Stanford University. He received his MSc degree in
> Artificial Intelligence from the University of Edinburgh, his Diplom (MS)
> degree in Computer Science from Saarland University, and his PhD (Dr.-Ing.)
> from the Max-Planck-Institute for Informatics.
>
>
> In his research he looks at algorithmic problems that lie at the
> intersection of Computer Graphics, Computer Vision and Machine Learning,
> such as: static and dynamic 3D scene reconstruction, neural rendering and
> neural scene representations, marker-less motion and performance capture,
> virtual humans, virtual and augmented reality, computer animation,
> intrinsic video and inverse rendering, computational videography, machine
> learning for graphics and vision, new sensors for 3D acquisition, as well
> as image- and physically-based rendering. He is also interested in using
> reconstruction techniques for human computer interaction. For his work, he
> received several awards, including the Otto Hahn Medal of the Max-Planck
> Society in 2007, the EUROGRAPHICS Young Researcher Award in 2009, the
> German Pattern Recognition Award 2012, and the Karl Heinz Beckurts Award in
> 2017, and the EUROGRAPHICS Outstanding Technical Contributions Award in
> 2020. He received two ERC grants, an ERC Starting Grant in 2013 and an ERC
> Consolidator Grant in 2017.
>
> *Zoom Link: *
> https://cmu.zoom.us/j/92289727241?pwd=a1VJUSsrYjJhblY1VEgzZFdMM2pLZz09
>
> Best,
> Shaojie Bai (MLD)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/ai-seminar-announce/attachments/20211214/8e188f21/attachment.html>