/Natural-Language-Process

DNN based implementation for different NLP tasks

Primary LanguagePython

This project is for implement different neural network based on theano:

  1. Semantic Role labeling based on open-source ATIS dataset;(Finished)
  2. Sentiment Classification;
  3. Machine Translation;
  4. Transition-based Paser;
  5. NLG;
  6. Chinese segmentation

Different metwork architecture will be supported:

  1. CNN
  2. RNN
    2.1 LSTM cell
    2.2 GRU cell(Finished)
  3. Sequence2Sequence
  4. GAN
  5. DQN

Different machinism will be supported:

  1. Different optimization algorithm:
    1.1 SGD(Finished)
    1.2 Adam
    1.3 Adagram
    1.4 Adadelta
  2. Attention machinism;
  3. Beam Search;
  4. Different initialize methods(Finished);

Usage:
cd src/nn
sh run.sh