Tea Time Talks are back for another year. This summer lecture series, presented by Amii and the RLAI Lab at the University of Alberta, give researchers the chance to discuss early-stage ideas and prospective research. Join us for another series of informal 20-minute talks where AI leaders discuss the future of machine learning research.
Abstract:
Identifying and learning from important events are crucial for data-efficient learning. However, current deep RL methods mostly use randomized experience selection.
The best-known alternative, prioritizing samples according to TD error, improves sample efficiency in tabular prediction problems but fails to improve upon random sample selection when used with neural networks.
I will describe the random and prioritized experience replay in RL, then outline an experiment in a chain MDP where prioritization should allow faster value propagation. The mixed results suggest we need better methods for experience selection in deep RL.
4 окт 2024
