A Learning-Exploring Method to Generate Diverse Paraphrases with Multi-Objective Deep Reinforcement Learning.
Mingtong LiuErguang YangDeyi XiongYujie ZhangYao MengChangjian HuJinan XuYufeng ChenPublished in: COLING (2020)
Keyphrases
- reinforcement learning
- multi objective
- learning process
- learning algorithm
- significant improvement
- optimization algorithm
- unsupervised learning
- clustering method
- generation method
- dynamic programming
- prior knowledge
- function approximators
- function approximation
- detection method
- neural network
- evolutionary algorithm
- probabilistic model
- temporal difference learning
- multi agent
- learning scheme
- learning problems
- fitted q iteration
- high accuracy
- supervised learning
- semi supervised
- information extraction
- support vector machine
- cost function
- natural language
- bayesian networks