Transfer of Deep Reactive Policies for MDP Planning.
Aniket BajpaiSankalp Garg MausamPublished in: CoRR (2018)
Keyphrases
- optimal policy
- markov decision problems
- markov decision processes
- markov decision process
- macro actions
- partially observable markov decision processes
- reactive planning
- state space
- reactive agents
- reward function
- planning problems
- reinforcement learning
- initial state
- finite state
- probabilistic planning
- partially observable
- stochastic domains
- planning under uncertainty
- dynamic programming algorithms
- decision theoretic planning
- decision theoretic
- heuristic search
- linear programming
- policy iteration
- dynamic programming
- average reward
- decision problems
- admissible heuristics
- infinite horizon
- knowledge transfer
- discounted reward
- ai planning
- policy search
- temporally extended
- utility function
- total reward
- bayesian reinforcement learning
- blocks world
- heuristic function
- long run
- domain independent