Characterizing the Action-Generalization Gap in Deep Q-Learning.
Zhiyuan ZhouCameron AllenKavosh AsadiGeorge KonidarisPublished in: CoRR (2022)
Keyphrases
- action selection
- reinforcement learning
- state action
- function approximation
- cooperative
- state space
- deep learning
- learning algorithm
- action space
- dynamic programming
- support vector
- multi agent
- data sets
- evaluation function
- learning rate
- knowledge base
- temporal difference learning
- reinforcement learning methods
- continuous state
- agent learns