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