Partially Observable Reinforcement Learning for Dialog-based Interactive Recommendation.
Yaxiong WuCraig MacdonaldIadh OunisPublished in: RecSys (2021)
Keyphrases
- partially observable
- reinforcement learning
- state space
- markov decision processes
- partial observability
- partially observable domains
- dynamical systems
- decision problems
- partially observable environments
- recommender systems
- mixed initiative
- markov decision problems
- hidden state
- infinite horizon
- partial observations
- reinforcement learning algorithms
- belief space
- reward function
- multi agent
- action models
- collaborative filtering
- function approximation
- learning algorithm
- belief state
- optimal policy
- temporal difference
- model free
- partially observable markov decision process
- machine learning
- dynamic systems
- objective function
- decision making