Timezone: »

Exponential Family Model-Based Reinforcement Learning via Score Matching
Gene Li · Junbo Li · Nathan Srebro · Zhaoran Wang · Zhuoran Yang
Event URL: https://openreview.net/forum?id=9GqTPzU1va »
We propose a optimistic model-based algorithm, dubbed SMRL, for finite-horizon episodic reinforcement learning (RL) when the transition model is specified by exponential family distributions with $d$ parameters and the reward is bounded and known. SMRL uses score matching, an unnormalized density estimation technique that enables efficient estimation of the model parameter by ridge regression. SMRL achieves $\tilde O(d\sqrt{H^3T})$ regret, where $H$ is the length of each episode and $T$ is the total number of interactions.

Author Information

Gene Li (Toyota Technological Institute at Chicago)
Junbo Li (UC Santa Cruz)
Nathan Srebro (University of Toronto)
Zhaoran Wang (Princeton University)
Zhuoran Yang (Princeton)

More from the Same Authors