Agenda

September 27

8:00 - 8:55 Breakfast and Check-in

8:55 - 9:00 Welcome and Introduction

9:00 - 10:00 Talk Session I, Session Chair - Elad Hazan

• Christos Papadimitriou - On biologically plausible artificial neural networks
• Phil Long + Hanie Sedghi - Size-free generalization bounds for convolutional neural networks
• Rina Panigrahy - How does the mind store information in memory?
• Raman Arora - Understanding the inductive bias due to dropout
• Vitaly Feldman - Does Learning Require Memorization? A Short Tale about a Long Tail

10:00 - 10:15 Break

10:15 - 11:15 Talk Session II, Session Chair - Yishay Mansour

• Leon Bottou - Learning Representations Using Causal Invariance
• Jieming Mao - On the Power of Interactivity in Local Differential Privacy
• Nisheeth Vishnoi - Physics-inspired Algorithms: Hamiltonian Monte Carlo for Sampling
• Sergei Vassilvitskii - Algorithms with Predictions
• Manish Purohit - Three Flavors of Predictions in Online Algorithms

11:15 - 11:30 Break

11:30 - 12:30 Talk Session III, Session Chair - Tim Roughgarden

• Claudio Gentile - On Active Learning with Zooming
• Steve Hanneke - Toward Optimal Agnostic Active Learning
• Karthik Sridharan - Towards Building Non-polarizing Recommendation Algorithms
• Brendan McMahan - Open Problems in Federated Learning
• Ananda Theertha Suresh - Agnostic Federated Learning

12:30 - 2:00 Lunch

2:00 - 3:00 Talk Session IV, Session Chair - Sasha Rakhlin

• Kunal Talwar - On the Error Resistance of Hinge Loss Minimization
• Pranjal Awasthi - On robustness to adversarial examples and polynomial optimization
• Andrew Cotter - On Making Stochastic Classifiers Deterministic
• Shivani Agarwal - Multiclass Learning with General Losses: What is the Right Output Coding and Decoding?
• Thodoris Lykouris - Adversarial robustness for stochastic bandit learning

3:00 - 3:15 Break

3:15 - 4:15 Talk Session V, Session Chair - Robert Schapire

• Haipeng Luo - Beating Stochastic and Adversarial Semi-bandits Optimally and Simultaneously
• Alex Slivkins - Incentivized exploration
• Naman Agarwal + Karan Singh - Provably Robust Control
• Akshay Krishnamurthy - Provably Efficient Reinforcement Learning with Rich Observations
• Jon Schneider + Renato Paes Leme - Contextual Search via Intrinsic Volumes

4:15 - 4:30 Break

4:30 - 5:45 Talk Session VI, Session Chair - Tony Jebara

• Satyen Kale - Escaping Saddle Points with Adaptive Gradient Methods
• Dylan Foster - The Complexity of Making the Gradient Small in Stochastic Optimization
• Francesco Orabana - Momentum-Based Variance Reduction in Non-Convex SGD
• Vahab Mirrokni - Adaptivity in Submodular Optimization
• Rong Ge - Explaining Landscape Connectivity of Low-cost Solutions for Multilayer Nets
• Stefanie Jegelka - How Powerful are Graph Neural Networks?

5:45 - 6:00 Walk to Toro for Evening Social

6:00 - 9:00 Evening Social at Toro, 85 10th Ave. NY, NY 10011