Share Reading Reading "ACL17 DeepSRL"
一个新的SRL模型,4层双向LSTM + highway + RNN dropout • 05 July 2017
A new srl model, 4 layer bi-lstm + highway + RNN dropout.
Model
- applying recent advances in training deep recurrent neural networks such as highway connections (Srivastava et al., 2015)
- RNN-dropouts (Gal and Ghahramani, 2016).
- using an A* decoding algorithm (Lewis and Steedman, 2014; Lee et al., 2016) to enforce structural consistency at prediction time without adding more complexity to the training process.
Analysis is complete
- What is the model good at and what kinds of mistakes does it make?
- How well do LSTMs model global structural consistency, despite conditionally indepen- dent tagging decisions?
- Is our model implicitly learning syntax, and could explicitly modeling syntax still help?