A Three-Stage Training Method for Abstractive Summarization.
Linqing ShiDanyang LiuGongshen LiuPublished in: Aust. J. Intell. Inf. Process. Syst. (2019)
Keyphrases
- high accuracy
- training samples
- training algorithm
- error rate
- neural network
- training phase
- detection method
- probabilistic model
- learning stage
- feed forward neural networks
- classification accuracy
- experimental evaluation
- prior knowledge
- pairwise
- semi supervised
- model selection
- hidden markov models
- cost function
- synthetic data
- high precision
- training process
- preprocessing
- objective function