Generative Sequence Models for Sequential Decision Making

Explore the intersection of generative sequence models and sequential decision making in this seminar presented by Aditya Grover, Assistant Professor of Computer Science at UCLA. Delve into a framework that abstracts sequential decision making as a generative sequence modeling problem, leveraging the Transformer architecture and advances in language modeling. Learn how this approach enables learning from large offline datasets, uncertainty-guided online exploration, and generalization across multiple tasks. Discover how the framework performs on various benchmarks, from continuous control to game playing, matching or exceeding state-of-the-art algorithms. Gain insights into efficient machine learning approaches for probabilistic reasoning under limited supervision, with applications in climate science and sustainable energy.