ELLIS Program "Interactive Learning and Interventional Representations" Workshop
21 June 2021 - 21 June 2021 Workshop
The ELLIS program for Interactive Learning and Interventional Representations (ILIR) decided to create a platform for the students and postdocs of program members, to present their work to all program members and to each other. To this end, all Fellows and Scholars of the program were invited to nominate speakers and topics, and we assembled a program. The initial plan was for a single workshop, but since there were so many high quality nominations, it was decided to have two workshops (June 14 and 21, 3pm-6pm CET via zoom).
The program, printed below, is representative of the different ELLIS program themes in ILIR. For questions please contact the program directors (Andreas Krause, Nicolo Cesa-Bianchi, Bernhard Schölkopf).
Monday June 14 2021
15:00-15:10 Introduction
15:10-15:30 Luigi Gresele: Independent mechanisms analysis, a new concept?
15:30-15:50 Jalal Etesami: Learning Randomly Translated Multivariate Hawkes Processes
15:50-16:10 Chaochao Lu: Nonlinear invariant risk minimization
16:10-16:30 Esther Derman: Reinforcement Learning with Execution Delay
16:40-17:00 Gellert Weisz: RL is hard, even with low-dimensional linear structure -- but all is not lost
17:00-17:20 Aviv Rosenberg: Oracle-Efficient Reinforcement Learning in Factored MDPs with Unknown Structure
17:20-17:40 Dhruva Tirumala: Behavior Priors for efficient RL
17:40-18:00 Julia Olkhovskaya: Efficient algorithms for adversarial contextual bandits and MDPs
Monday June 21 2021
15:00-15:20 Siqi Liu: Scalable Multi-Agent Learning in Real-World Games?
15:20-15:40 Pier Giuseppe Sessa: Efficient Learning in Games with Structure
15:40-16:00 Vitaly Kurin: Incompatible Multitask Deep Reinforcement Learning
16:00-16:20 Pierre Laforgue: Multitask Online Mirror Descent
16:30-16:50 Giulia Denevi: Conditional meta-learning
16:50-17:10 Fabio Cermelli: Towards open world semantic segmentation across domains
17:10-17:30 So Takao: Geometric deep learning for data-driven weather forecasting
17:30-17:50 Joshua Robinson: Improving contrastive learning with hard examples
17:50-18:00 Wrap-up
