Foundations of Safe Learning

An IBM Research AI Week Workshop

September 20, 2019 | Cambridge, MA

Foundations of Safe Learning

As deep learning moves from the lab into real-world applications, ensuring the correctness and robustness of deep neural networks becomes a paramount concern. Specifying what it means for a neural network to behave correctly is a challenging problem, especially for classifiers and generative models. Verifying that deep networks meet these specifications is computationally challenging. Recent demonstrations of brittleness in deep learning – including adversarial examples and RL agents that learn pathological control policies – have motivated new computationally tractable approaches toward both specifying and verifying salient properties of neural networks. This workshop will bring together academics, industry researchers, and practitioners to delve into the current state of safe learning. Relevant topics include robustness evaluation/verification, safe reinforcement learning, fairness, robustness against model compression, interpretability, and deep learning with invariance and equivariance constraints.

This workshop is part of IBM AI Research Week.

Registration

Registration is free but required for attendance. You may register at any time, including the day of the event.

Tentative Schedule

Welcome

8:45 am to 9:00 am   Registration
9:00 am to 9:05 am   Opening Remarks
Nathan Fulton

Session #1

9:05 am to 9:35 am   Towards AI You Can Rely On
Aleksander Madry
9:35 am to 10:05 am   Yanzhi Wang
10:05 am to 10:35 am   Secure Learning in Adversarial Environments
Bo Li

Coffee Break

10:35 am to 10:50 am   Coffee and light refreshments

Session #2

10:50 am to 11:20 am   Stefanie Jegelka
11:20 am to 11:50 am   A Formal Method's Perspective to AI Safety: Promises and Challenges
Wenchao Li
11:50 am to 12:20 pm   Can Deep Learning Models Be Trusted?
Luca Daniel
12:20 pm to 12:25 pm   Closing Remarks
Sijia Liu

Venue

The workshop will take place in the Samberg Conference Center (map below) in room DR 3.

Speakers

Aleksander Madry
(Massachusetts Institute of Technology)
Yanzhi Wang
(Northeastern University)
Bo Li
(University of Illinois at Urbana-Champaign)
Stefanie Jegelka
(Massachusetts Institute of Technology)
Wenchao Li
(Boston University)
Luca Daniel
(Massachusetts Institute of Technology)

Contact

Questions should be directed to nathan@ibm.com

Workshop Organizers

Nathan Fulton
(MIT-IBM)
Quanfu Fan
(MIT-IBM)
Sijia Liu
(MIT-IBM)