Softmax Deep Double Deterministic Policy Gradients

TL;DR

This paper introduces a novel softmax operator for value estimation in continuous control reinforcement learning, addressing overestimation bias and improving performance over existing methods like DDPG and TD3.

Contribution

It proposes the SD2 and SD3 algorithms that incorporate the softmax operator, providing a new approach to mitigate bias in actor-critic algorithms for continuous control.

Findings

SD3 outperforms state-of-the-art methods in continuous control tasks.

The softmax operator smooths the optimization landscape, aiding learning.

Theoretical analysis reveals benefits of the softmax operator in actor-critic algorithms.

Abstract

A widely-used actor-critic reinforcement learning algorithm for continuous control, Deep Deterministic Policy Gradients (DDPG), suffers from the overestimation problem, which can negatively affect the performance. Although the state-of-the-art Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm mitigates the overestimation issue, it can lead to a large underestimation bias. In this paper, we propose to use the Boltzmann softmax operator for value function estimation in continuous control. We first theoretically analyze the softmax operator in continuous action space. Then, we uncover an important property of the softmax operator in actor-critic algorithms, i.e., it helps to smooth the optimization landscape, which sheds new light on the benefits of the operator. We also design two new algorithms, Softmax Deep Deterministic Policy Gradients (SD2) and Softmax Deep Double Deterministic Policy Gradients (SD3), by building the softmax operator upon single and double estimators, which can effectively improve the overestimation and underestimation bias. We conduct extensive experiments on challenging continuous control tasks, and results show that SD3 outperforms state-of-the-art methods.