[1911.02140] Fully Parameterized Quantile Function for Distributional Reinforcement Learning

Distributional Reinforcement Learning (RL) differs from traditional RL in that, rather than the expectation of total returns, it estimates distributions and has achieved state-of-the-art performance on Atari Games. The key challenge in practical distributional RL algorithms lies in how to parameterize estimated distributions so as to better approximate the true continuous distribution. Existing distributional RL algorithms parameterize either the probability side or the return value side of the distribution function, leaving the other side uniformly fixed as in C51, QR-DQN or randomly sampled as in IQN. In this paper, we propose fully parameterized quantile function that parameterizes both the quantile fraction axis (i.e., the x-axis) and the value axis (i.e., y-axis) for distributional RL. Our algorithm contains a fraction proposal network that generates a discrete set of quantile fractions and a quantile value network that gives corresponding quantile values. The two networks are joi

4 mentions: @mooopan@pm_girl@helioRocha_
Date: 2019/11/07 11:20

Referring Tweets

@mooopan Fully Parameterized Quantile Function for Distributional Reinforcement Learning t.co/ef1tsYRm5f New SoTA on Atari under non-distributed settings from Microsoft Research

Related Entries

Read more [DL輪読会]Deep Reinforcement Learning that Matters
0 users, 0 mentions 2018/05/04 16:46
Read more ML Platform Meetup: Infra for Contextual Bandits and Reinforcement Learning
3 users, 12 mentions 2019/10/18 19:55
Read more Seminar Report: The 25th STAIR Lab AI Seminar: Dr. Eiji Uchibe, “Imitation Learning based on a Combi...
0 users, 1 mentions 2019/11/03 05:20
Read more RL— Introduction to Deep Reinforcement Learning – Jonathan Hui – Medium
0 users, 0 mentions 2018/10/15 01:08
Read more [1909.11830] Improving SAT Solver Heuristics with Graph Networks and Reinforcement Learning
0 users, 5 mentions 2019/09/29 17:18