Spatiotemporally Constrained Action Space Attacks on Deep Reinforcement Learning Agents.
Xian Yeow LeeSambit GhadaiKai Liang TanChinmay HegdeSoumik SarkarPublished in: AAAI (2020)
Keyphrases
- action space
- reinforcement learning agents
- reinforcement learning
- state space
- state abstraction
- markov decision processes
- dynamic environments
- single agent
- real valued
- multi agent
- heuristic search
- function approximation
- action selection
- optimal policy
- multi agent environments
- machine learning
- markov decision problems
- reinforcement learning methods
- function approximators
- control problems
- markov decision process
- dynamic programming
- learning algorithm
- transfer learning
- reinforcement learning algorithms
- feature selection
- temporal difference
- data mining
- supervised learning