Preserving the Privacy of Reward Functions in MDPs through Deception.
Shashank Reddy ChirraPradeep VarakanthamPraveen ParuchuriPublished in: CoRR (2024)
Keyphrases
- reward function
- markov decision processes
- reinforcement learning
- state space
- optimal policy
- markov decision process
- reinforcement learning algorithms
- policy search
- inverse reinforcement learning
- privacy preserving
- multiple agents
- partially observable
- personal information
- transition probabilities
- markov decision problems
- dynamic programming
- state variables
- state action
- average reward
- factored mdps
- privacy concerns
- infinite horizon
- sensitive information
- markov chain
- average cost
- data mining
- policy iteration
- action space
- finite state
- control policies
- complex systems
- generative model
- learning algorithm