Fwd: Aleksander Madry on Why Do ML Models Fail?

Tue Feb 16 10:23:52 EST 2021

This will be of interest to those of us whose work touches reliability and
trustworthiness of ML.

Artur

---------- Forwarded message ---------
From: C3.ai Digital Transformation Institute <info at c3dti.ai>
Date: Tue, Feb 16, 2021 at 10:16 AM
Subject: Aleksander Madry on Why Do ML Models Fail?
To: <awd at cs.cmu.edu>

Aleksander Madry on Why Do ML Models Fail?
[image: C3DTI Colloquium on Digital Transformation Science]
<https://c3dti.us8.list-manage.com/track/click?u=aae4610086c8b3a8646b00cbe&id=0cf721f153&e=b5ed94badf>

The Colloquium on Digital Transformation is a series of weekly online talks
on how artificial intelligence, machine learning, and big data can lead to
scientific breakthroughs with large-scale societal benefit. The spring 2021
series focuses largely on COVID-19 mitigation research.

*See details of upcoming talks here
<https://c3dti.us8.list-manage.com/track/click?u=aae4610086c8b3a8646b00cbe&id=bc0e6d7ac4&e=b5ed94badf>
and
note we have the same Zoom Webinar registration
<https://c3dti.us8.list-manage.com/track/click?u=aae4610086c8b3a8646b00cbe&id=3e6c706cbc&e=b5ed94badf>
link
for all forthcoming talks*
 *Why Do ML Models Fail?*

*February 18, 1 pm PT/4 pm ET*

*Aleksander Madry*

*Professor of Computer Science Massachusetts Institute of Technology*

<https://c3dti.us8.list-manage.com/track/click?u=aae4610086c8b3a8646b00cbe&id=228810bb87&e=b5ed94badf>

Our current machine learning (ML) models achieve impressive performance on
many benchmark tasks. Yet, these models remain remarkably brittle,
susceptible to manipulation and, more broadly, often behave in ways that
are unpredictable to users. Why is this the case? In this talk, we identify
human-ML misalignment as a chief cause of this behavior. We then take an
end-to-end look at the current ML training paradigm and pinpoint some of
the roots of this misalignment. We discuss how current pipelines for
dataset creation, model training, and system evaluation give rise to
unintuitive behavior and widespread vulnerability. Finally, we conclude by
outlining possible approaches towards alleviating these deficiencies.

Aleksander Madry
<https://c3dti.us8.list-manage.com/track/click?u=aae4610086c8b3a8646b00cbe&id=182acacc3a&e=b5ed94badf>
is a Professor of Computer Science at MIT and leads the MIT Center for
Deployable Machine Learning. His research interests span algorithms,
continuous optimization, science of deep learning, and understanding
machine learning from a robustness and deployability perspectives.
Aleksander's work has been recognized with a number of awards, including an
NSF CAREER Award, an Alfred P. Sloan Research Fellowship, an ACM Doctoral
Dissertation Award Honorable Mention, and Presburger Award. He received his
PhD from MIT in 2011 and, prior to joining the MIT faculty, he spent time
at Microsoft Research New England and on the faculty of EPFL.
For those who missed last week's C3.ai DTI Colloquium, you can catch it on
our YouTube channel Colloquium Playlist
<https://c3dti.us8.list-manage.com/track/click?u=aae4610086c8b3a8646b00cbe&id=3aee0a74e5&e=b5ed94badf>,
where videos of all weekly talks are available.
<https://c3dti.us8.list-manage.com/track/click?u=aae4610086c8b3a8646b00cbe&id=9880bf9944&e=b5ed94badf>
C3.ai DTI Colloquium: February 11, 2021

*Scoring Drugs: Small Molecule Drug Discovery for COVID-19 using
Physics-Inspired Machine Learning*

Teresa Head-Gordon
Chancellor’s Professor, Department of Chemistry, Chemical
and Biomolecular Engineering, and Bioengineering
University of California, Berkeley

The rapid spread of SARS-CoV-2 has spurred the scientific world into action
for therapeutics to help minimize fatalities from COVID-19. Molecular
modeling is combating the current global pandemic through the traditional
process of drug discovery, but the slow turnaround time for identifying
leads for antiviral drugs, analyzing structural effects of genetic
variation in the evolving virus, and targeting relevant virus-host protein
interactions is still a great limitation during an acute crisis. The first
component of drug discovery - the structure of potential drugs and the
target proteins - has driven functional insight into biology ever since
Watson, Crick, Franklin, and Wilkins solved the structure of DNA. What
could we do with structural models of host and virus proteins and small
molecule therapeutics? We can further enrich structure with dynamics for
discovery of new surface sites exposed by fluctuations to bind drugs and
peptide therapeutics not revealed by a static structural model. These
“cryptic” binding sites offer new leads in drug discovery but will only
yield fruit if they can be assessed rapidly for binding affinity for new
small molecule drugs. We offer physics-inspired data-driven models to: 1)
extend the chemical space of new drugs beyond those available; 2) create
reliable scoring functions to evaluate drug binding affinities to cryptic
binding sites of COVID-19 targets; 3) accelerate computation of binding
affinities by training machine learning models; and 4) closing the loop of
design and evaluation to bias the distribution of new drug candidates
towards desired metrics enabled by the C3 AI Suite.
*About the C3.ai Digital Transformation Institute*

Established in March 2020 by C3 AI, Microsoft, and leading universities,
the C3.ai Digital Transformation Institute is a research consortium
dedicated to accelerating the socioeconomic benefits of artificial
intelligence. The Institute engages the world’s leading scientists to
conduct research and train practitioners in the new Science of Digital
Transformation, which operates at the intersection of artificial
intelligence, machine learning, cloud computing, internet of things, big
data analytics, organizational behavior, public policy, and ethics. The
nine C3.ai Digital Transformation Institute consortium member universities
and laboratories are: University of Illinois at Urbana-Champaign;
University of California, Berkeley; Carnegie Mellon University; Lawrence
Berkeley National Laboratory; Massachusetts Institute of Technology;
National Center for Supercomputing Applications at University of Illinois
at Urbana-Champaign; Princeton University; Stanford
University; and University of Chicago. Learn more at C3DTI.ai
<https://c3dti.us8.list-manage.com/track/click?u=aae4610086c8b3a8646b00cbe&id=6c87501aef&e=b5ed94badf>
.
[image: Twitter]
<https://c3dti.us8.list-manage.com/track/click?u=aae4610086c8b3a8646b00cbe&id=139cb3c2de&e=b5ed94badf>
[image: LinkedIn]
<https://c3dti.us8.list-manage.com/track/click?u=aae4610086c8b3a8646b00cbe&id=a2f874999b&e=b5ed94badf>
[image: Facebook]
<https://c3dti.us8.list-manage.com/track/click?u=aae4610086c8b3a8646b00cbe&id=df5e82568e&e=b5ed94badf>
[image: YouTube]
<https://c3dti.us8.list-manage.com/track/click?u=aae4610086c8b3a8646b00cbe&id=c8230f1630&e=b5ed94badf>
[image: Website]
<https://c3dti.us8.list-manage.com/track/click?u=aae4610086c8b3a8646b00cbe&id=b47714cc8b&e=b5ed94badf>
[image: Email] <info at c3dti.ai>

*C3.ai Digital Transformation Institute @ Berkeley*
University of California, Berkeley
750 Sutardja Dai Hall, MC 1764
Berkeley, California 94720-1764

*C3.ai Digital Transformation Institute @ Illinois*
University of Illinois at Urbana-Champaign
1205 W. Clark Street, MC-257, Room 1008
Urbana, Illinois 61801
*Copyright © 2021 C3.ai Digital Transformation Institute, All rights
reserved.*
You are receiving this email because you opted in via our website.

Want to change how you receive these emails?
You can update your preferences
<https://c3dti.us8.list-manage.com/profile?u=aae4610086c8b3a8646b00cbe&id=2aac18199b&e=b5ed94badf&c=55d1d77794>
or
unsubscribe from this list
<https://c3dti.us8.list-manage.com/unsubscribe?u=aae4610086c8b3a8646b00cbe&id=2aac18199b&e=b5ed94badf&c=55d1d77794>
.

[image: Email Marketing Powered by Mailchimp]
<http://www.mailchimp.com/email-referral/?utm_source=freemium_newsletter&utm_medium=email&utm_campaign=referral_marketing&aid=aae4610086c8b3a8646b00cbe&afl=1>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20210216/93c71050/attachment-0001.html>