From awd at cs.cmu.edu  Mon Nov  4 10:54:03 2024
From: awd at cs.cmu.edu (Artur Dubrawski)
Date: Mon, 4 Nov 2024 10:54:03 -0500
Subject: Fwd: 24-25 AY RI Departmental Fellowships
In-Reply-To: <4718B07E-F9A7-4E20-9A8C-84B74FB7B379@andrew.cmu.edu>
References: <4718B07E-F9A7-4E20-9A8C-84B74FB7B379@andrew.cmu.edu>
Message-ID: <CAJvAoysuygi_tXBqfo6Mmqhy-WAP+wikPLRaa9q+y_PSaw9UOA@mail.gmail.com>

Please join me in congratulating Ceci on winning the Uber Presidential
Fellowship Award!

Cheers!
Artur

---------- Forwarded message ---------
From: George Kantor <gkantor at andrew.cmu.edu>
Date: Mon, Nov 4, 2024 at 10:33?AM
Subject: 24-25 AY RI Departmental Fellowships
To: RI People <ri-people at andrew.cmu.edu>


Please join me in congratulating to this year?s recipients of the RI
Departmental PhD Fellowships:

RI Presidential Fellowship: Tairan He

Uber Presidential Fellowships (3 awards):
Samantha Speer
Iqui-Balam Heredia Marin
Cecilia Morales

Robotics Vision Fellowship: Ayush Jain

Red Whittaker Endowed Field Robotics: Paulo Fisch

Quality of Life Tech Center Student Research Fund: Zulekha Karachiwalla

Paul Tompkins Memorial Fellowship: Abigail Breitfeld

Alan Guisewite Memorial Fellowship: Philip (Yizhou) Huang


Thank you to all of the faculty who submitted applications on behalf of
their students, and an extra special thank you to the review committee
(Yannis, Jiaoyang, Basti, Cam).

All fellowships are for the 2024-2025 academic year, and must be spent by
June 30th 2025. If you are advising one of the recipients, Becky will be in
touch with additional instructions.

Thanks,
-George

George Kantor
Research Professor / Associate Director of Education
The Robotics Institute
Carnegie Mellon University
gkantor at andrew.cmu.edu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20241104/46b7a909/attachment.html>

From jeff4 at andrew.cmu.edu  Tue Nov  5 12:12:31 2024
From: jeff4 at andrew.cmu.edu (Jeff Schneider)
Date: Tue, 5 Nov 2024 12:12:31 -0500
Subject: Fwd: Thesis Proposal - November 5, 2024 - Youngseog Chung - Methods
 for Calibrated Uncertainty Quantification and Understanding its Utility
In-Reply-To: <5597591b-6d87-4b88-ab60-7955e80fdc60@andrew.cmu.edu>
References: <5597591b-6d87-4b88-ab60-7955e80fdc60@andrew.cmu.edu>
Message-ID: <1bb7cfec-9342-4656-4bdc-b9cdb2b84db0@andrew.cmu.edu>

Hi Everyone,

You're invited to see Youngseog's thesis proposal starting at 1pm today!

Jeff.



-------- Forwarded Message --------
Subject: 	Thesis Proposal - November 5, 2024 - Youngseog Chung - Methods 
for Calibrated Uncertainty Quantification and Understanding its Utility
Date: 	Wed, 30 Oct 2024 05:27:41 -0400
From: 	Diane L Stidle <stidle at andrew.cmu.edu>
To: 	ml-seminar at cs.cmu.edu <ML-SEMINAR at CS.CMU.EDU>, jsnoek at google.com



*/Thesis Proposal/*

Date: November 5, 2024
Time: 1:00pm (EDT)
Place: GHC 4405
Speaker: Youngseog Chung

*Thesis Title:*
Methods for Calibrated Uncertainty Quantification and Understanding its 
Utility
*
*
*Abstract:*
As machine learning models have become more capable of dealing with 
complex data, they have been entrusted with an increasing array of 
predictive tasks. However, with growing reliance on model predictions, 
being able to assess whether a given model prediction is reliable has 
become equally important. Uncertainty quantification (UQ) plays a 
critical role in this context by providing a measure of confidence in a 
model's predictions, and the quantified uncertainty is considered 
correct if it is calibrated. In this proposal, I address the problem of 
optimizing for calibration, especially with regression models which 
output a distribution over continuous-valued outputs. In my initial 
work, I propose a collection of methods and techniques to train a 
quantile model end-to-end with differentiable loss functions that 
optimize directly for the calibration of the predictive quantiles. This 
work?falls under a class of pre-hoc methods which aim to improve 
calibration during the training of the model and distinguishes itself 
from the relatively richer line of work in post-hoc calibration, which 
aim to calibrate a pre-trained predictive model. Afterwards, I introduce 
a method to feasibly extend the notion of calibration to 
multi-dimensional distributions and describe a post-hoc calibration (or 
recalibration) algorithm. I further discuss how distributional 
predictions are utilized in applications such as decision-making tasks 
or model-based reinforcement learning and point out that each 
application setting requires different qualities for the distributional 
prediction.
In light of this observation, I propose several research directions 
which study applications of using distributional predictions. In 
particular, I propose re-investigating proper scoring rules as a tool 
for eliciting good/useful behavior from distributional predictions in a 
pre-hoc manner.
*
Thesis Committee:*
Jeff Schneider (Chair)
Aarti Singh
Zico Kolter
Jasper Snoek (Google Deepmind)
*
*
*Link to the draft document: *
https://youngseogchung.github.io/docs/thesis_proposal.pdf

*Zoom meeting link:*
https://cmu.zoom.us/my/youngseog.chung


From awd at cs.cmu.edu  Wed Nov  6 09:40:23 2024
From: awd at cs.cmu.edu (Artur Dubrawski)
Date: Wed, 6 Nov 2024 09:40:23 -0500
Subject: Fwd: Special Seminar with Chris Golias
In-Reply-To: <3F958CA7-EF65-4778-857B-2D39938CB27F@andrew.cmu.edu>
References: <3F958CA7-EF65-4778-857B-2D39938CB27F@andrew.cmu.edu>
Message-ID: <CAJvAoyuOeBL-y0WTnga1UHax=32i2i8UTU-w73Zn65WZr-0fdg@mail.gmail.com>

This may be relevant to all Autonians who touch user experience aspects of
AI.

Artur

---------- Forwarded message ---------
From: Carolyn Buzzelli-Stumpf <cbstumpf at andrew.cmu.edu>
Date: Wed, Nov 6, 2024 at 9:28?AM
Subject: Special Seminar with Chris Golias
To: <scs-all at cs.cmu.edu>
Cc: Angelica Stowers <astowers at andrew.cmu.edu>


*An Invitation to all SCS:*

Special Seminar with Chris Golias
*November 11, 2024 at 10:00 am*
Newell Simon Hall 4305
Human- Computer Interaction Institute
Faculty Host: Dan Saffer

*Title:* Epistemology for UX Researchers: How to choose the right research
method, every time.

*Abstract*: The purpose of user experience research (UXR) is to create
reliable, actionable knowledge about a product, journey or user group.
Across public, private, industry, start-up and academic contexts, job
postings emphasize that even junior UXRs should demonstrate the ability to
discern the proper method or methods to address key business questions and
execute rigorous studies that generate the kinds of knowledge necessary for
making informed decisions. This seminar seeks to demystify the UXR method
selection process by providing students with a systematic way to think
about the knowledge creation processes and mechanisms behind common UXR
techniques like usability tests, in-depth interviews, diary studies and
surveys. I will begin with an introduction to epistemology, the study of
knowledge, during which I will orient students to three primary
epistemological frameworks-rationalism, empiricism and existentialism. I
then move on to distinguish deductive and inductive reasoning, empirical
versus attitudinal research methods and how to identify the object of
research, all while providing examples from my research practice in
industry. The demonstration concludes with group time for students to
select appropriate methods for several curated product research scenarios.

*Bio*: Christopher Golias, Ph.D., is a technology anthropologist, currently
Senior Researcher for internationalization at Google's Gemini, who has
conducted applied anthropological research across various areas including
retail, healthcare, indigenous rights, substance use, ecommerce,
governance, machine learning, localization and information technology. He
holds a Ph.D. in Anthropology from the University of Pennsylvania.




[image: Profile.jpeg]



*Zoom Link:*
https://cmu.zoom.us/j/99590424526?pwd=Xl6YgsjhtsEeJ1HpExkTC8N7EgbLna.1
Meeting ID: 995 9042 4526
Passcode: 072547

This presentation is intended for members of the SCS community and relevant
CMU stakeholders. Do not forward this link to anyone outside of the SCS/CMU
community.


Thank you,
Carolyn

- - -
Carolyn Buzzelli-Stumpf
Project Coordinator/Assistant to:
Professor Brad Myers, Department Head
Professor and Associate Dean, Jodi Forlizzi
Human-Computer Interaction Institute | The School of Computer Science
Carnegie Mellon University | 5000 Forbes Ave | Newell Simon Hall 3515
o: 412-268-1001 | cbstumpf at andrew.cmu.edu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20241106/0ce534f5/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Profile.jpeg
Type: image/jpeg
Size: 22461 bytes
Desc: not available
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20241106/0ce534f5/attachment-0001.jpeg>

From awd at cs.cmu.edu  Mon Nov 11 16:51:38 2024
From: awd at cs.cmu.edu (Artur Dubrawski)
Date: Tue, 12 Nov 2024 06:51:38 +0900
Subject: Fwd: FW: S3D Distinguished Speakers Series: Satish Chandra - Tuesday
 11/12
In-Reply-To: <05ab8526c76976cd2ccc79529eb9642d@mail.gmail.com>
References: <7abf22ca37c1aa9f7c07489845c14f39@mail.gmail.com>
 <78a3296d28dee5ce91732198a9694d91@mail.gmail.com>
 <05ab8526c76976cd2ccc79529eb9642d@mail.gmail.com>
Message-ID: <CAJvAoyuQKh3nAc0a0XL1oryejzt8-mPeb2hY3qZ3A6A0AegOLg@mail.gmail.com>

I know this will be of interest to many of us.

Cheers
Artur

---------- Forwarded message ---------
From: Linda Campbell <lv2c at andrew.cmu.edu>
Date: Tue, Nov 12, 2024, 2:24?AM
Subject: FW: S3D Distinguished Speakers Series: Satish Chandra - Tuesday
11/12
To: <ml-seminar at cs.cmu.edu>


S3D Distinguished Speakers Series



Tuesday, November 12, 2024

12:00 p.m. ? 1:15 p.m.   Google calendar
<https://calendar.google.com/calendar/event?action=TEMPLATE&tmeid=MXZsZHM5bHI3MGhjdjhpbmoxMThmZWdmbWIgYW5kcmV3LmNtdS5lZHVfbjBscGtkaWFyYTJhdWhqaXVjbmpsbmVoYzhAZw&tmsrc=andrew.cmu.edu_n0lpkdiara2auhjiucnjlnehc8%40group.calendar.google.com>

In person at TCS 358

Overflow/watch room TCS 460

Or online via Zoom:
https://cmu.zoom.us/j/98473836934?pwd=Sm1ZK0dzamF4eEZHNmlGbFFYK0RmQT09
*Lunch provided starting at 11:40 a.m.*



*Title:*  AI in Software Engineering at Google



*Speaker:*  Satish Chandra, Principal Engineer, Core Developer Group, Google



*Abstract:*  The availability of large code corpora, coupled with advances
in machine learning have made it possible to build software development
tools grounded in statistical principles, paving way for capabilities that
were not possible with technologies that relied on program semantics alone.
In the last couple of years, large language models have taken the world by
storm, and have shown remarkable ability to create, explain, critique and
debug code.   In this talk, I will provide an overview of this area both
from a research as well as technology point of view. I?ll describe how at
Google we have been working on weaving AI capabilities in developer
workflows, how we collect data, how we prioritize our work, and the impact
that this work is showing. I will discuss how model quality and user
experience interact in interesting ways that have a material implication on
the success of such tools. I will share some thoughts on how the field is
evolving from its erstwhile focus on code completion to higher-level
tasks.  Finally, I will discuss the role of benchmarks in our community,
and what we can do to push the state of the art together at a higher pace.



*Bio:*  Satish Chandra is a researcher at Google, where he applies machine
learning techniques to improve developer productivity. His work has spanned
many areas of programming languages and software engineering, including
program analysis, type systems, software synthesis, bug finding and repair,
software testing and, of course, application of AI to software development.
His research has been widely published in leading conferences in his
field.  Satish Chandra obtained a PhD from the University of
Wisconsin-Madison, and a B.Tech from the Indian Institute of
Technology-Kanpur, both in computer science. He is an ACM Distinguished
Scientist and an elected member of WG 2.4.



*Upcoming S3D Seminar Series Talks*



January 29:  Nick Feamster*

February 5:  Vladimir Filkov*

April 30:  Danah Boyd*



Links available to recordings are here.  Usually published a couple of days
after the talk.

https://s3d.cmu.edu/events/distinguished.html
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20241112/c308dda8/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.png
Type: image/png
Size: 2585446 bytes
Desc: not available
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20241112/c308dda8/attachment-0001.png>

From awd at cs.cmu.edu  Tue Nov 12 18:32:13 2024
From: awd at cs.cmu.edu (Artur Dubrawski)
Date: Wed, 13 Nov 2024 08:32:13 +0900
Subject: Fwd: RI Ph.D. Thesis Proposal: Shuli Jiang
In-Reply-To: <CAPfXyxKPtNxAaLUQGfCNanL3KfN5iwrC=e_1d3sRhL+opi8BqQ@mail.gmail.com>
References: <CAPfXyxKPtNxAaLUQGfCNanL3KfN5iwrC=e_1d3sRhL+opi8BqQ@mail.gmail.com>
Message-ID: <CAJvAoyu5J4dNMbUNn0mxbUtbN6hREpkvws3L3w=Kju7stNdzgQ@mail.gmail.com>

Of interest to everyone into DAI and coming from a former Autonian.

Artur

---------- Forwarded message ---------
From: Suzanne Muth <lyonsmuth at cmu.edu>
Date: Wed, Nov 13, 2024, 7:23?AM
Subject: RI Ph.D. Thesis Proposal: Shuli Jiang
To: RI People <ri-people at andrew.cmu.edu>


*Date: *22 November 2024
*Time: *9:00 a.m. (ET)
*Location: *NSH 4305
*Zoom Link: *
https://cmu.zoom.us/j/97463407099?pwd=Tp69Yms7hu7kOkgA1jBiQLQPhTyxDe.1
<https://www.google.com/url?q=https://cmu.zoom.us/j/97463407099?pwd%3DTp69Yms7hu7kOkgA1jBiQLQPhTyxDe.1&sa=D&source=calendar&ust=1731745749580401&usg=AOvVaw2uNS6oDx-js-n2WM8Ssyd5>
*Type:* Ph.D. Thesis Proposal
*Who:* Shuli Jiang
*Title:* Communication Efficient and Differentially Private Optimization

*Abstract:*
In recent years, the integration of communication efficiency and
differential privacy in distributed optimization has gained significant
attention, motivated by large-scale applications such as Federated Learning
(FL), where both data privacy and efficient communication are critical.
This thesis explores the development of novel techniques to address these
challenges, with a focus on distributed mean estimation, differentially
private prediction, and private optimization for empirical risk
minimization.

The first part of this work addresses communication-efficient distributed
vector mean estimation, an essential subroutine in distributed optimization
and FL. We propose the Rand-Proj-Spatial family estimator which utilizes
cross-client correlation to reduce the estimation error under fixed
communication cost, by projecting client vectors into a random subspace
using a Subsampled Randomized Hadamard Transform (SRHT). This approach
captures cross-client correlation more effectively, demonstrating
substantial performance gains over conventional sparsification techniques
in various distributed optimization tasks.

The second part of this work focuses on maximizing the privacy-utility
trade-offs in differentially private prediction through majority
ensembling. We introduce the Data-dependent Randomized Response Majority
(DaRRM) framework, which generalizes all private majority ensembling
algorithms through a data-dependent noise function. Based on DaRRM, we
propose a computationally tractable optimization procedure for maximizing
utility under a fixed privacy loss. Empirical results demonstrate DaRRM?s
effectiveness in private label ensembling for image classification, showing
significant utility improvements over existing baselines.

The third part of this work investigates differentially private
optimization in solving empirical risk minimization using shuffled gradient
methods. Unlike conventional private optimizers such as DP-SGD, which
benefits from privacy amplification by subsampling, shuffled gradient
methods face unique challenges in privacy and convergence. We develop a
theoretical framework for analyzing Incremental Gradient (IG) methods, the
most basic form of shuffled gradient methods, that enables noise injection
for privacy and the use of surrogate objectives, introducing a new
dissimilarity metric to measure the difference between true and surrogate
objectives. Leveraging privacy amplification by iteration, we establish the
first empirical excess risk bound for differentially private IG (DP-IG),
and show how interleaving public data in training can further improve
privacy-convergence trade-offs in DP-IG.

Finally, we introduce two proposed works along the line of differentially
private optimization. First, we aim to extend our theoretical framework to
analyze Shuffle Once (SO) and Random Reshuffling (RR), two practical
shuffled gradient methods beyond Incremental Gradient (IG) methods. This
will enable us to understand their private counterparts, DP-SO and DP-RR,
where privacy analysis is more complex due to a lack of understanding on
privacy amplification through shuffling. Second, we plan to extend our
framework from a local to a distributed or decentralized setting to analyze
convergence rates of distributed shuffled gradient methods in both private
and non-private contexts, while also investigating the impact of data
heterogeneity among clients on convergence in this setting.

*Thesis Committee Members:*
Gauri Joshi, Chair
Steven Wu
Zachary Manchester
Swanand Kadhe, IBM Research

A draft of the thesis proposal document can be found here
<https://11hifish.github.io/content/thesis_proposal.pdf>. (available after
Nov.18)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20241113/8d0970dd/attachment.html>

From awd at cs.cmu.edu  Sun Nov 17 15:49:57 2024
From: awd at cs.cmu.edu (Artur Dubrawski)
Date: Mon, 18 Nov 2024 05:49:57 +0900
Subject: Fwd: Guest Seminar @ ML4PDE Group Meeting Tuesday 11/19/24
In-Reply-To: <CAGUCJsPxsrShWJgVKQDE6MLHTYQ15RQ0hqbPr6qDsX=wn3-caQ@mail.gmail.com>
References: <CAGUCJsPxsrShWJgVKQDE6MLHTYQ15RQ0hqbPr6qDsX=wn3-caQ@mail.gmail.com>
Message-ID: <CAJvAoysXNkpL1vEC+ArQAH5s8vOVn=ALqY=OuVVbR4tq9v9Ddg@mail.gmail.com>

Fysa

---------- Forwarded message ---------
From: Nicholas Boffi <nboffi at andrew.cmu.edu>
Date: Mon, Nov 18, 2024, 4:11?AM
Subject: Guest Seminar @ ML4PDE Group Meeting Tuesday 11/19/24
To: <ml-all at cs.cmu.edu>


Hi All,

On *Tuesday 11/19* from *3-4pm* in *Wean Hall 7218*, we will have a guest
lecture from *Michalis Michaelides*, head of research at PhysicsX (
https://www.physicsx.ai). PhysicsX is a recent startup that aims to use
machine learning-based techniques for partial differential equations to
accelerate engineering design problems. Any students or faculty in MLD
broadly interested in this area are more than welcome to attend.

As usual, our persistent Zoom link is below.

Join Zoom Meeting
https://cmu.zoom.us/j/94390770161?pwd=Syl6Rx4JRf78cFuSiNhDoqzG7IaTXs.1

Meeting ID: 943 9077 0161
Passcode: 584858

Hope to see you there,
Nick

-- 
Nicholas M. Boffi, Assistant Professor
Department of Mathematical Sciences
Wean Hall 6121
https://nmboffi.github.io
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20241118/235338c7/attachment.html>

From awd at cs.cmu.edu  Tue Nov 19 18:30:42 2024
From: awd at cs.cmu.edu (Artur Dubrawski)
Date: Tue, 19 Nov 2024 18:30:42 -0500
Subject: Autonian companies rocking the boat in Tech 50 Awards
Message-ID: <CAJvAoyuG-POJ+wTQt2iX5jpgR2kvBMyb9oZs+RzB+PeKNM0oUw@mail.gmail.com>

Dear Autonians,

I am very pleased to let you all know that our own spinoff Marinus
Analytics won the Pittsburgh Technology Council's Innovator of the Year
award in the Solutions Provider ? Services category.

https://www.pghtech.org/news-and-publications/Tech50_2024_2

Also notable in the list of awardees is Lovelace AI, nominated in
the AI/ML/Robotics category. It is a startup company recently founded by
our fearless founder Andrew Moore.

Way to go Cara & Team and Andrew & Team!

Artur
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20241119/cae107c8/attachment.html>

From awd at cs.cmu.edu  Tue Nov 19 18:39:49 2024
From: awd at cs.cmu.edu (Artur Dubrawski)
Date: Tue, 19 Nov 2024 18:39:49 -0500
Subject: a new book by our prolific author
Message-ID: <CAJvAoytYgBMDp8k_vxR1dnRmp-xxDzFAvMb3dQaqhRiXNY5Ssw@mail.gmail.com>

Autonians,

I am very pleased to announce a new book authored by our own Jeremy Kubica:

https://nostarch.com/graph-algorithms-fun-way

After a series of enormously successful books introducing computer science
to kids, this new publication, as well as the one that precedes it, aims at
more adult audiences. It serves as a neat intro to graph algorithms. Looks
like a great read!

Congrats Jeremy, and please do not slow down!

Cheers,
Artur

PS Here is a link to Jeremy's amazon.com portfolio:
https://www.amazon.com/stores/Jeremy-Kubica/author/B00957OB3Q?ref=ap_rdr&isDramIntegrated=true&shoppingPortalEnabled=true
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20241119/28be34d4/attachment.html>

From awd at cs.cmu.edu  Wed Nov 20 10:35:05 2024
From: awd at cs.cmu.edu (Artur Dubrawski)
Date: Wed, 20 Nov 2024 10:35:05 -0500
Subject: Fwd: SCS Katayanagi Distinguished Lecture: Thursday, November 21 -
 4:30 pm - GHC 4401
In-Reply-To: <CAMXnxCynwMZCZxq0FLoS+QBV7gMAMqF+FPRf-cKoGhpC9hyYRQ@mail.gmail.com>
References: <CAMXnxCynwMZCZxq0FLoS+QBV7gMAMqF+FPRf-cKoGhpC9hyYRQ@mail.gmail.com>
Message-ID: <CAJvAoytOxG1SzkWy68zRmn1hgRPSJ22aPN2hopg+W-kbmv1ebQ@mail.gmail.com>

This is highly relevant to at least a few of us.

Artur

---------- Forwarded message ---------
From: Catherine Copetas <copetas at cs.cmu.edu>
Date: Fri, Nov 8, 2024, 2:29?PM
Subject: SCS Katayanagi Distinguished Lecture: Thursday, November 21 - 4:30
pm - GHC 4401
To: <scs-all at cs.cmu.edu>



*Please join us for the next...*

*SCS Katayanagi Distinguished Lecture*
Thursday,* 21 November 2024*
*4:30 pm *
Rashid Auditorium, *Gates Hillman 4401*


*as we welcome*

*NOAM BROWN <https://noambrown.github.io/>*
Research Scientist
OpenAI
Learning to Reason with LLMs
Large language models (LLMs) have demonstrated remarkable capabilities in
generating coherent text and completing various natural language tasks.
Nevertheless, their ability to perform complex, general reasoning has
remained limited. In this talk, I will describe OpenAI's new o1 model, an
LLM trained via reinforcement learning to generate a hidden chain of
thought before its response. We have found that the performance of o1
consistently improves with more reinforcement learning compute and with
more inference compute. o1 surpasses previous state-of-the-art models in a
variety of benchmarks that require reasoning, including mathematics
competitions, programming contests, and advanced science question sets. I
will discuss the implications of scaling this paradigm even further.

?

*Noam Brown <https://noambrown.github.io/>* is a research scientist at
OpenAI investigating reasoning and multi-agent AI. He co-created Libratus
<https://en.wikipedia.org/wiki/Libratus> and Pluribus
<https://www.nytimes.com/2019/07/11/science/poker-robot-ai-artificial-intelligence.html>,
the first AIs to defeat top humans in two-player no-limit poker and
multiplayer no-limit poker, respectively, and Cicero, the first AI to
achieve human-level performance in the natural language strategy game
Diplomacy. He has received the Marvin Minsky Medal for Outstanding
Achievements in AI, was named one of *MIT Tech Review's 35 Innovators Under
35*, and his work on Pluribus was named by *Science* as one of the top 10
scientific breakthroughs of 2019. Noam received his Ph.D. *(CS)* from
Carnegie Mellon University.


*About the Lecture:  The Katayanagi Lectures recognize the best and the
brightest in the field of computer science and are presented by the School
of Computer Science at Carnegie Mellon University in cooperation with the
Tokyo University of Technology (TUT). The lectures recognize both senior
and junior talent.  The series were established through a gift from
Japanese entrepreneur and education advocate, Mr. Koh Katayanagi, who
founded TUT and other technical institutions in Japan over many multiple
decades.*

Questions <scs-dls at cs.cmu.edu>?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20241120/89f2f76c/attachment.html>

From awd at cs.cmu.edu  Mon Nov 25 11:08:18 2024
From: awd at cs.cmu.edu (Artur Dubrawski)
Date: Mon, 25 Nov 2024 11:08:18 -0500
Subject: Cristian Challu's thesis nominated for the AAAI/ACM SIGAI
 Dissertation Award!
Message-ID: <CAJvAoyso6htUdKk1KHA8X6L6n8HxMfYL-o9BxxOb7xLgV6hVxA@mail.gmail.com>

Dear Autonians,

I am very pleased to let you know that Cristian's thesis is the sole entry
representing CMU in this year's AAAI/ACM SIGAI Doctoral Dissertation Award
competition.

It is already a great achievement to be nominated from a very highly
selective community that produces a couple of dozens of high quality
AI-oriented dissertations each year, but we of course keep our fingers
crossed for the award committee making the correct decision as well. I am
hopeful given how impactful Cristian's PhD work is just looking at the
number of citations of his AAAI 2023 paper on NHITS (counting 381 at Google
Scholar now), but of course the competitors ought to be pretty strong as
well.

Please join me in cheering for Cristian and his thesis!

Artur
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20241125/06d2c23d/attachment.html>

From awd at cs.cmu.edu  Mon Nov 25 11:53:16 2024
From: awd at cs.cmu.edu (Artur Dubrawski)
Date: Mon, 25 Nov 2024 11:53:16 -0500
Subject: Cristian Challu's thesis nominated for the AAAI/ACM SIGAI
 Dissertation Award!
In-Reply-To: <CAJvAoyso6htUdKk1KHA8X6L6n8HxMfYL-o9BxxOb7xLgV6hVxA@mail.gmail.com>
References: <CAJvAoyso6htUdKk1KHA8X6L6n8HxMfYL-o9BxxOb7xLgV6hVxA@mail.gmail.com>
Message-ID: <CAJvAoysOQ8aSpKwfczZF2MDB8KGshj-kQkdXEVB7MTLSN3XfvA@mail.gmail.com>

Ah and of course I forgot to mention that the famous NHITS paper is a
brilliant example of Auton collaboration. It has two equal first authors in
Cristian and Kin Gutierrez Olivares!

Artur

On Mon, Nov 25, 2024, 11:08?AM Artur Dubrawski <awd at cs.cmu.edu> wrote:

> Dear Autonians,
>
> I am very pleased to let you know that Cristian's thesis is the sole entry
> representing CMU in this year's AAAI/ACM SIGAI Doctoral Dissertation Award
> competition.
>
> It is already a great achievement to be nominated from a very highly
> selective community that produces a couple of dozens of high quality
> AI-oriented dissertations each year, but we of course keep our fingers
> crossed for the award committee making the correct decision as well. I am
> hopeful given how impactful Cristian's PhD work is just looking at the
> number of citations of his AAAI 2023 paper on NHITS (counting 381 at Google
> Scholar now), but of course the competitors ought to be pretty strong as
> well.
>
> Please join me in cheering for Cristian and his thesis!
>
> Artur
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20241125/75860b0c/attachment.html>

From cchallu at andrew.cmu.edu  Tue Nov 26 14:10:53 2024
From: cchallu at andrew.cmu.edu (Cristian Challu)
Date: Tue, 26 Nov 2024 11:10:53 -0800
Subject: Cristian Challu's thesis nominated for the AAAI/ACM SIGAI
 Dissertation Award!
In-Reply-To: <CAJvAoyso6htUdKk1KHA8X6L6n8HxMfYL-o9BxxOb7xLgV6hVxA@mail.gmail.com>
References: <CAJvAoyso6htUdKk1KHA8X6L6n8HxMfYL-o9BxxOb7xLgV6hVxA@mail.gmail.com>
Message-ID: <CAEwPDOGF3EcFMu4_4HaLYqQuGMADpGvvKQqqEraDRFstTXmESw@mail.gmail.com>

Thanks, Professor, for sharing this news with the lab and for your kind
words. It?s truly an honor to represent CMU in such a prestigious
competition.

This recognition wouldn?t have been possible without the collaborations
with the brilliant minds at the Auton Lab!

I look forward to visiting CMU soon to share our latest research and
connect with everyone.

Best,
Cristian

On Mon, Nov 25, 2024 at 8:10?AM Artur Dubrawski <awd at cs.cmu.edu> wrote:

> Dear Autonians,
>
> I am very pleased to let you know that Cristian's thesis is the sole entry
> representing CMU in this year's AAAI/ACM SIGAI Doctoral Dissertation Award
> competition.
>
> It is already a great achievement to be nominated from a very highly
> selective community that produces a couple of dozens of high quality
> AI-oriented dissertations each year, but we of course keep our fingers
> crossed for the award committee making the correct decision as well. I am
> hopeful given how impactful Cristian's PhD work is just looking at the
> number of citations of his AAAI 2023 paper on NHITS (counting 381 at Google
> Scholar now), but of course the competitors ought to be pretty strong as
> well.
>
> Please join me in cheering for Cristian and his thesis!
>
> Artur
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20241126/81c2e637/attachment.html>