Cohere For AI - Guest Speaker: Akifumi Wachi, Sr Research Scientist

Join our Reinforcement Learning group as they host Akifumi Wachi to present "Safe RL."

Speaker's Bio: Akifumi Wachi is a Chief Research Scientist at LY Corporation. His research interests lie primarily in reinforcement learning and span the entire theory-to-application spectrum from fundamental advances to deployment in real-world systems. He is especially interested in how a policy should and can be trained and deployed in safety-critical problems. HP: https://akifumi-wachi-4.github.io/website/ Description: Safety is an essential problem when deploying reinforcement learning (RL) to real applications. Consequently, safe RL emerges as a fundamental yet powerful paradigm for optimizing an agent's policy from experimental data, which is typically formulated as a problem where the expected cumulative reward is maximized under safety constraints. The talk will first give an overview of safe RL and then discuss a few algorithms we have recently proposed including SNO-MDP (https://proceedings.mlr.press/v119/wachi20a.html) and MASE (https://openreview.net/forum?id=dQLsvKNwZC).

Add event to calendar

Apple Google Office 365 Outlook Outlook.com Yahoo