Research Seminar "Machine Learning Theory"

This is the research seminar by Ulrike's group.

When and where

Each thursday 14:00 - 15:00. Until further notice, all meetings take place virtually! To participate in the virtual meetings, please drop an email to Sebastian.

What

Most sessions take place in form of a reading group: everybody reads the assigned paper before the meeting. Then we jointly discuss the paper in the meeting. Sometimes we also have talks by guests or members of the group.

Who

Everybody who is interested in machine learning theory: Students, PhD students and researchers of the University of Tübingen. We do not mind people dropping in and out depending on whether they find the current session interesting or not.

Upcoming meetings

14.1.2021 no meeting

Explainability

21.1.2021 Paper discussion (who?), A Unified Approach to Interpreting Model Predictions, NeurIPS 2017.
28.1.2021 no meeting (IMPRS interviews)
4.2.2021 Paper discussion (Sebastian), Anchors: High-Precision Model-Agnostic Explanations, AAAI 2018.
11.2.2021 Paper discussion (who?), Fairwashing Explanations with Off-Manifold Detergent, ICML 2020.

Causality

18.2.2021 Paper discussion (Luca), Shapley Flow: A Graph-based Approach to Interpreting Model Predictions, arXiv, 2020.
25.2.2021 Paper discussion (Sascha), Generative causal explanations of black-box classifiers, NeurIPS, 2020.
4.3.2021 Paper discussion (Leena), Conditional variance penalties and domain shift robustness., Machine Learning, 2020. Drop Section 5.3 and following.

Past meetings

Listed here.

Suggested papers for future meetings

Feel free to make suggestions!
If you do, please (i) try to select short conference papers rather than 40-page-journal papers; (ii) please put your name when entering suggestions; it does not mean that you need to present it, but then we can judge where it comes from;

Paper discussion (who?)Random deep neural networks are biased towards simple functions, NeurIPS 2019.
Ulrike: Underspecification Presents Challenges for Credibility in Modern Machine Learning, Arxiv nov 2020.
On the Convergence of Stochastic Gradient Descent with Low-Rank Projections for Convex Low-Rank Matrix Problems Dan Garber , COLT 2020
(?) On the Utility of Learning about Humans for Human-AI Coordination, NeurIPS 2019.
(Ulrike) Strength from Weakness: Fast Learning Using Weak Supervision Joshua Robinson, Stefanie Jegelka, Suvrit Sra link
(Sascha): The Power of Batching in Multiple Hypothesis Testing, Tijana Zrnic, Daniel Jiang, Aaditya Ramdas, Michael Jordan; PLMR, 2020.
Auditing ML Models for Individual Bias and Unfairness Songkai Xue, Mikhail Yurochkin, Yuekai Sun ; PMLR 108:4552-4562 http://proceedings.mlr.press/v108/xue20a/xue20a.pdf
Fast and Bayes-consistent nearest neighbors Klim Efremenko, Aryeh Kontorovich, Moshe Noivirt ; Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics, PMLR 108:1276-1286, 2020. http://proceedings.mlr.press/v108/efremenko20a/efremenko20a.pdf
Envy-free classification, NeurIPS 2019.
On the Fairness of Disentangled Representations, NeurIPS 2019.
(Leena)To Understand Deep Learning We Need to Understand Kernel Learning Mikhail Belkin, Siyuan Ma, Soumik Mandal, ICML 2018
(Leena) Neural Tangent Kernel:Convergence and Generalization in Neural Networks Arthur Jacot, Franck Gabriel, Clement Hongler, NeurIPS 2018
Rademacher Complexity for Adversarially Robust Generalization ICML 2019
(Michael) Two-Player Games for Efficient Non-Convex Constrained Optimization ALT 2019
(Michael) Balancing Competing Objectives with Noisy Data - Score-Based Classifier for Welfare-Aware Machine Learning ICML 2020
(Michael) Minimax Pareto Fairness: A Multi Objective Perspective ICML 2020