Published: 2019, Last Modified: 12 May 2023ICML 2019Readers: Everyone
Abstract:Q-learning methods are a common class of algorithms used in reinforcement learning (RL). However, their behavior with function approximation, especially with neural networks, is poorly understood t...