Music & AI Plus A Geometric Perspective on Reinforcement Learning with Pablo Samuel Castro

Today we're joined by Pablo Samuel Castro, Staff Research Software Developer at Google. Pablo, whose research is mainly focused on reinforcement learning, and I caught up at NeurIPS last month. We cover a lot of ground in our conversation, including his love for music, and how that has guided his work on the Lyric AI project, which holds the goal of building an artificial intelligence assistant designed to help musicians with create original lyrics. We talk through the evolution of language models, the influence of other research in this area, and Pablo also gives us quite a few references, which you can find below. We also spend time discussing a few of his other NeurIPS submissions, including "A Geometric Perspective on Optimal Representations for Reinforcement Learning," which "proposes a new perspective on representation learning in reinforcement learning based on geometric properties of the space of value functions" and "Estimating Policy Functions in Payments Systems using Deep Reinforcement Learning," which aims to show the applicability of Deep Reinforcement Learning to estimate best-response functions in real-world situations.