Ubiquitous Distributed Deep Reinforcement Learning at the Edge: Analyzing Byzantine Agents in Discrete Action Spaces.
Wenshuai ZhaoJorge Peña QueraltaLi QingqingTomi WesterlundPublished in: CoRR (2020)
Keyphrases
- action space
- reinforcement learning
- multi agent
- single agent
- continuous state
- continuous state spaces
- state space
- continuous action
- action selection
- state and action spaces
- fault tolerant
- markov decision processes
- multi agent systems
- cooperative
- multiple agents
- distributed systems
- real valued
- mobile agents
- control policies
- policy search
- reinforcement learning methods
- skill learning
- learning agent
- function approximation
- function approximators
- state action
- decision problems
- reinforcement learning algorithms
- stochastic processes
- markov decision process
- dynamic environments
- temporal difference
- markov chain
- model free
- decision making
- robot navigation
- stochastic games
- state dependent
- machine learning
- finite number
- optimal policy
- learning algorithm