Timezone: »

 
Target Entropy Annealing for Discrete Soft Actor-Critic
Yaosheng Xu · Dailin Hu · Litian Liang · Stephen McAleer · Pieter Abbeel · Roy Fox

@
in Deep Reinforcement Learning »
Event URL: https://openreview.net/forum?id=jJKzGBBQiZu »
Soft Actor-Critic (SAC) is considered the state-of-the-art algorithm in continuous action space settings. It uses the maximum entropy framework for efficiency and stability, and applies a heuristic temperature Lagrange term to tune the temperature $\alpha$, which determines how "soft" the policy should be. It is counter-intuitive that empirical evidence shows SAC does not perform well in discrete domains. In this paper we investigate the possible explanations for this phenomenon and propose Target Entropy Scheduled SAC (TES-SAC), an annealing method for the target entropy parameter applied on SAC. Target entropy is a constant in the temperature Lagrange term and represents the target policy entropy in discrete SAC. We compare our method on Atari 2600 games with different constant target entropy SAC, and analyze on how our scheduling affects SAC.

Author Information

Yaosheng Xu (University of California, Irvine)
Dailin Hu (UC Irvine)
Litian Liang (University of California, Irvine)
Stephen McAleer (UC Irvine)
Pieter Abbeel (UC Berkeley & Covariant)

Pieter Abbeel is Professor and Director of the Robot Learning Lab at UC Berkeley [2008- ], Co-Director of the Berkeley AI Research (BAIR) Lab, Co-Founder of covariant.ai [2017- ], Co-Founder of Gradescope [2014- ], Advisor to OpenAI, Founding Faculty Partner AI@TheHouse venture fund, Advisor to many AI/Robotics start-ups. He works in machine learning and robotics. In particular his research focuses on making robots learn from people (apprenticeship learning), how to make robots learn through their own trial and error (reinforcement learning), and how to speed up skill acquisition through learning-to-learn (meta-learning). His robots have learned advanced helicopter aerobatics, knot-tying, basic assembly, organizing laundry, locomotion, and vision-based robotic manipulation. He has won numerous awards, including best paper awards at ICML, NIPS and ICRA, early career awards from NSF, Darpa, ONR, AFOSR, Sloan, TR35, IEEE, and the Presidential Early Career Award for Scientists and Engineers (PECASE). Pieter's work is frequently featured in the popular press, including New York Times, BBC, Bloomberg, Wall Street Journal, Wired, Forbes, Tech Review, NPR.

Roy Fox (UC Irvine)
Roy Fox

[Roy Fox](royf.org) is an Assistant Professor and director of the Intelligent Dynamics Lab at the Department of Computer Science at UCI. His research interests include theory and applications of reinforcement learning, algorithmic game theory, information theory, and robotics. His current research focuses on structure, exploration, and optimization in deep reinforcement learning and imitation learning of virtual and physical agents and multi-agent systems. He was previously a postdoc at UC Berkeley, where he developed algorithms and systems that interact with humans to learn structured control policies for robotics and program synthesis.

More from the Same Authors