Reinforcement Learning with Dynamic Boltzmann Softmax Updates
This addresses a convergence problem in reinforcement learning for value estimation, offering a method that enhances performance in Atari games, though it is incremental as it builds on existing softmax operators.
The paper tackles the convergence issue of the Boltzmann softmax operator in reinforcement learning by proposing a dynamic Boltzmann softmax (DBS) operator, which improves value function estimation and outperforms DQN in 40 out of 49 Atari games.
Value function estimation is an important task in reinforcement learning, i.e., prediction. The Boltzmann softmax operator is a natural value estimator and can provide several benefits. However, it does not satisfy the non-expansion property, and its direct use may fail to converge even in value iteration. In this paper, we propose to update the value function with dynamic Boltzmann softmax (DBS) operator, which has good convergence property in the setting of planning and learning. Experimental results on GridWorld show that the DBS operator enables better estimation of the value function, which rectifies the convergence issue of the softmax operator. Finally, we propose the DBS-DQN algorithm by applying dynamic Boltzmann softmax updates in deep Q-network, which outperforms DQN substantially in 40 out of 49 Atari games.