Bohan Chen, "Recurrent neural network encoder-decoder model adaptation for text paraphrase"

2016/11/28 18:37 に 後藤太一 が投稿

In this report,  he does adaption approach to RNN encoder-decoder translation system for the purpose of resolving lack of big parallel data.
Experiment using Shakespeare corpus  show that self mirroring system works well, but adaptation do not work.
Future works is testing conventional adaption methods and loss functions.