Deep Reinforcement Learning with a Combinatorial Action Space for Predicting Popular Reddit Threads.
Ji HeMari OstendorfXiaodong HeJianshu ChenJianfeng GaoLihong LiLi DengPublished in: EMNLP (2016)
Keyphrases
- action space
- reinforcement learning
- state space
- markov decision processes
- state and action spaces
- continuous state
- real valued
- continuous state spaces
- reinforcement learning methods
- action selection
- control policies
- stochastic processes
- function approximators
- markov decision process
- function approximation
- state action
- multi agent
- machine learning
- model free
- heuristic search
- learning algorithm
- temporal difference
- reinforcement learning algorithms
- control problems
- single agent
- finite state
- transfer learning
- optimal policy
- sufficient conditions
- higher order