Discourse-Aware Neural Rewards for Coherent Text Generation.
Antoine BosselutAsli CelikyilmazXiaodong HeJianfeng GaoPo-Sen HuangYejin ChoiPublished in: NAACL-HLT (2018)
Keyphrases
- text generation
- natural language generation
- natural language
- network architecture
- reinforcement learning
- theorem prover
- bio inspired
- neural network
- bandit problems
- neural fuzzy
- markov decision processes
- multi armed bandits
- multiarmed bandit
- long term and short term
- associative memory
- natural language processing
- dialogue system
- ground truth
- artificial intelligence
- nonlinear predictive control
- learning rules
- credit assignment
- discourse structure
- hebbian learning
- inference rules
- machine learning