Trends in Reinforcement Learning with Pablo Samuel Castro - #443
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
00:00
Exploring Contrastive Loss in Reinforcement Learning
This chapter examines the role of contrastive loss within reinforcement learning, emphasizing its use in learning state representations. It highlights the significance of bisimulation metrics and their relevance in simplifying complex systems through state aggregation. Furthermore, the discussion extends to innovative techniques that integrate biological insights and new methods for approximating bisimulation metrics using deep neural networks.
Transcript
Play full episode
