An end-to-end inverse reinforcement learning by a boosting approach with relative entropy.
Tao ZhangYing LiuMaxwell HwangKao-Shing HwangChunyan MaJing ChengPublished in: Inf. Sci. (2020)
Keyphrases
- end to end
- relative entropy
- inverse reinforcement learning
- information theoretic
- preference elicitation
- information theory
- bregman divergences
- log likelihood
- boosting algorithms
- mutual information
- covariance matrix
- reward function
- cost sensitive
- learning algorithm
- feature selection
- temporal difference
- maximum entropy
- machine learning
- base classifiers
- graphical models
- support vector machine
- hidden markov models