Mask-Attention A3C: Visual Explanation of Action-State Value in Deep Reinforcement Learning.
Hidenori ItayaTsubasa HirakawaTakayoshi YamashitaHironobu FujiyoshiKomei SugiuraPublished in: IEEE Access (2024)
Keyphrases
- reinforcement learning
- state space
- action space
- selective attention
- multi agent
- action selection
- initial state
- action sequences
- sensory inputs
- state variables
- visual information
- state action
- state transition
- dynamic programming
- computer vision
- partially observable domains
- state abstraction
- reward shaping
- machine learning
- visual field
- visual perception
- model free
- function approximation
- supervised learning
- digital images
- low level