( events)   Timezone: »  
Program Highlights »
Sat Dec 14 08:00 AM -- 06:00 PM (PST) @ West Ballroom B
Machine Learning with Guarantees
Ben London · Gintare Karolina Dziugaite · Daniel Roy · Thorsten Joachims · Aleksander Madry · John Shawe-Taylor

Workshop Home Page

As adoption of machine learning grows in high-stakes application areas (e.g., industry, government and health care), so does the need for guarantees: how accurate a learned model will be; whether its predictions will be fair; whether it will divulge information about individuals; or whether it is vulnerable to adversarial attacks. Many of these questions involve unknown or intractable quantities (e.g., risk, regret or posterior likelihood) and complex constraints (e.g., differential privacy, fairness, and adversarial robustness). Thus, learning algorithms are often designed to yield (and optimize) bounds on the quantities of interest. Beyond providing guarantees, these bounds also shed light on black-box machine learning systems.

Classical examples include structural risk minimization (Vapnik, 1991) and support vector machines (Cristianini & Shawe-Taylor, 2000), while more recent examples include non-vacuous risk bounds for neural networks (Dziugaite & Roy, 2017, 2018), algorithms that optimize both the weights and structure of a neural network (Cortes, 2017), counterfactual risk minimization for learning from logged bandit feedback (Swaminathan & Joachims, 2015; London & Sandler, 2019), robustness to adversarial attacks (Schmidt et al., 2018; Wong & Kolter, 2018), differentially private learning (Dwork et al., 2006, Chaudhuri et al., 2011), and algorithms that ensure fairness (Dwork et al., 2012).

This one-day workshop will bring together researchers in both theoretical and applied machine learning, across areas such as statistical learning theory, adversarial learning, fairness and privacy, to discuss the problem of obtaining performance guarantees and algorithms to optimize them. The program will include invited and contributed talks, poster sessions and a panel discussion. We particularly welcome contributions describing fundamentally new problems, novel learning principles, creative bound optimization techniques, and empirical studies of theoretical findings.

Welcome Address (Talk)
Ben London
Tengyu Ma, "Designing Explicit Regularizers for Deep Models" (Invited Talk)
Tengyu Ma
Vatsal Sharan, "Sample Amplification: Increasing Dataset Size even when Learning is Impossible" (Contributed Talk)
Vatsal Sharan
Break / Poster Session 1
Antonia Marcu, Yao-Yuan Yang, Pascale Gourdeau, Chen Zhu, Thodoris Lykouris, Jianfeng Chi, Mark Kozdoba, Arjun Nitin Bhagoji, Xiaoxia (Shirley) Wu, Jay Nandy, Michael T Smith, Bingyang Wen, Yuege (Gail) Xie, Konstantinos Pitas, Suprosanna Shit, Maksym Andriushchenko, Dingli Yu, Gaël Letarte, Misha Khodak, Hussein Mozannar, Chara Podimata, James Foulds, Yizhen Wang, Huishuai Zhang, Ondrej Kuzelka, Alexander Levine, Nan Lu, Zakaria Mhammedi, Paul Viallard, Diana Cai, Lovedeep Gondara, James Lucas, Yasaman Mahdaviyeh, Aristide Baratin, Rishi Bommasani, Alessandro Barp, Andrew Ilyas, Kaiwen Wu, Jens Behrmann, Omar Rivasplata, Amir Nazemi, Aditi Raghunathan, William Stephenson, Sahil Singla, Akhil Gupta, YooJung Choi, Yannic Kilcher, Clare Lyle, Edoardo Manino, Andrew Bennett, Zhi Xu, Niladri Chatterji, Emre Barut, Flavien Prost, Rodrigo Toro Icarte, Arno Blaas, Charlie Yun, Sahin Lale, YiDing Jiang, Tharun Medini, Ashkan Rezaei, Alexander Meinke, Stephen Mell, Gary Kazantsev, Shivam Garg, Anu Sinha, Vishnu Lokhande, Geovani Rizk, Han Zhao, Aditya Kumar Akash, Jikai Hou, Ali Ghodsi, Matthias Hein, Tyler Sypherd, Yichen Yang, Anastasia Pentina, Pierre Gillot, Antoine Ledent, Guy Gur-Ari, Noah MacAulay, Tianzong Zhang
Mehryar Mohri, "Learning with Sample-Dependent Hypothesis Sets" (Invited Talk)
Mehryar Mohri
James Lucas, "Information-theoretic limitations on novel task generalization" (Contributed Talk)
James Lucas
Lunch Break
Soheil Feizi, "Certifiable Defenses against Adversarial Attacks" (Invited Talk)
Soheil Feizi
Maksym Andriushchenko, "Provably Robust Boosted Decision Stumps and Trees against Adversarial Attacks" (Contributed Talk)
Maksym Andriushchenko
Coffee Break / Poster Session 2
Aaron Roth, "Average Individual Fairness" (Invited Talk)
Aaron Roth
Hussein Mozannar, "Fair Learning with Private Data" (Contributed Talk)
Hussein Mozannar
Emma Brünskill, "Some Theory RL Challenges Inspired by Education" (Invited Talk)
Emma Brunskill
Discussion Panel