Pseudo Dyna-Q: A Reinforcement Learning Framework for Interactive Recommendation.
Lixin ZouLong XiaPan DuZhuo ZhangTing BaiWeidong LiuJian-Yun NieDawei YinPublished in: WSDM (2020)
Keyphrases
- reinforcement learning
- main contribution
- database
- real time
- machine learning
- learning process
- temporal difference
- reinforcement learning methods
- function approximation
- user friendly
- markov decision processes
- virtual reality
- lightweight
- user interaction
- probabilistic model
- dynamic programming
- recommender systems
- case study