Deep Radial-Basis Value Functions for Continuous Control
This addresses the problem of efficient continuous control in RL for researchers and practitioners, offering a novel method that improves performance over existing approaches.
The paper tackles the challenge of finding optimal actions for continuous control in reinforcement learning by introducing deep radial-basis value functions (RBVFs), which enable easy and accurate approximation of maximum action-values. The resulting RBF-DQN agent significantly outperforms value-function-only baselines and is competitive with state-of-the-art actor-critic algorithms.
A core operation in reinforcement learning (RL) is finding an action that is optimal with respect to a learned value function. This operation is often challenging when the learned value function takes continuous actions as input. We introduce deep radial-basis value functions (RBVFs): value functions learned using a deep network with a radial-basis function (RBF) output layer. We show that the maximum action-value with respect to a deep RBVF can be approximated easily and accurately. Moreover, deep RBVFs can represent any true value function owing to their support for universal function approximation. We extend the standard DQN algorithm to continuous control by endowing the agent with a deep RBVF. We show that the resultant agent, called RBF-DQN, significantly outperforms value-function-only baselines, and is competitive with state-of-the-art actor-critic algorithms.