30 November 2020 - Josiah Hanna Speaker: Josiah Hanna Title: Data Efficient Reinforcement Learning from Re-weighted and Simulated Data Abstract: Learning from interaction with the environment -- trying untested actions, observing successes and failures, and tying effects back to causes -- is one of the first capabilities thought of when considering intelligent agents. Reinforcement learning is the area of artificial intelligence research that has the goal of allowing autonomous agents to learn in this way. Despite many recent empirical successes, most modern reinforcement learning algorithms are still limited by the large amounts of experience required before useful skills are learned. Making reinforcement learning more data efficient would allow computers to autonomously solve complex tasks in dynamic environments such as those found in robotics, traffic management, or healthcare. In this talk I will describe recent work towards increasing the data efficiency of reinforcement learning algorithms. In the first part of the talk, I will introduce a novel re-weighting technique that allows RL agents to more efficiently use a finite set of samples. The key idea behind this technique is to use importance sampling to convert the empirical distribution of samples to the expected distribution, thus reducing sampling error in the observed samples. I will describe how this technique can be flexibly applied in reinforcement learning: leading to more efficient batch policy evaluation, policy gradient reinforcement learning, and batch value function learning. In the second part of the talk, I will describe recent advances in using RL with off-the-shelf physics simulators to improve bipedal locomotion on physical robots. Bio: Josiah Hanna is a post-doc in the School of Informatics at the University of Edinburgh working with Stefano Albrecht. Starting August 2021, he will be an assistant professor in the Department of Computer Sciences at the University of Wisconsin -- Madison. He received his Ph.D. in the Computer Science Department at the University of Texas at Austin advised by Peter Stone. Prior to attending UT Austin, he completed his B.S. in computer science and mathematics at the University of Kentucky advised by Judy Goldsmith. Josiah is a recipient of the NSF Graduate Research Fellowship and the IBM Ph.D. Fellowship. His research interests lie in artificial intelligence and machine learning, seeking to develop algorithms that allow autonomous agents to learn (efficiently) from experience. In particular, he studies reinforcement learning and methods to increase the data efficiency of reinforcement learning algorithms. Nov 30 2020 14.00 - 15.00 30 November 2020 - Josiah Hanna AIAI Seminar talk hosted by Josiah Hanna Online
30 November 2020 - Josiah Hanna Speaker: Josiah Hanna Title: Data Efficient Reinforcement Learning from Re-weighted and Simulated Data Abstract: Learning from interaction with the environment -- trying untested actions, observing successes and failures, and tying effects back to causes -- is one of the first capabilities thought of when considering intelligent agents. Reinforcement learning is the area of artificial intelligence research that has the goal of allowing autonomous agents to learn in this way. Despite many recent empirical successes, most modern reinforcement learning algorithms are still limited by the large amounts of experience required before useful skills are learned. Making reinforcement learning more data efficient would allow computers to autonomously solve complex tasks in dynamic environments such as those found in robotics, traffic management, or healthcare. In this talk I will describe recent work towards increasing the data efficiency of reinforcement learning algorithms. In the first part of the talk, I will introduce a novel re-weighting technique that allows RL agents to more efficiently use a finite set of samples. The key idea behind this technique is to use importance sampling to convert the empirical distribution of samples to the expected distribution, thus reducing sampling error in the observed samples. I will describe how this technique can be flexibly applied in reinforcement learning: leading to more efficient batch policy evaluation, policy gradient reinforcement learning, and batch value function learning. In the second part of the talk, I will describe recent advances in using RL with off-the-shelf physics simulators to improve bipedal locomotion on physical robots. Bio: Josiah Hanna is a post-doc in the School of Informatics at the University of Edinburgh working with Stefano Albrecht. Starting August 2021, he will be an assistant professor in the Department of Computer Sciences at the University of Wisconsin -- Madison. He received his Ph.D. in the Computer Science Department at the University of Texas at Austin advised by Peter Stone. Prior to attending UT Austin, he completed his B.S. in computer science and mathematics at the University of Kentucky advised by Judy Goldsmith. Josiah is a recipient of the NSF Graduate Research Fellowship and the IBM Ph.D. Fellowship. His research interests lie in artificial intelligence and machine learning, seeking to develop algorithms that allow autonomous agents to learn (efficiently) from experience. In particular, he studies reinforcement learning and methods to increase the data efficiency of reinforcement learning algorithms. Nov 30 2020 14.00 - 15.00 30 November 2020 - Josiah Hanna AIAI Seminar talk hosted by Josiah Hanna Online