/robsut-wrod-reocginiton

Robsut Wrod Reocginiton via semi-Character Recurrent Neural Network

Primary LanguagePython

Data and re-implemented scripts for 2017 AAAI paper, "Robsut Wrod Reocginiton via semi-Character Recurrent Neural Network"

Last updated: Oct 8th , 2017


This repository contains the re-implementation on the following paper:

@inproceedings{DBLP:conf/aaai/SakaguchiDPD17,
  author    = {Keisuke Sakaguchi and
               Kevin Duh and
               Matt Post and
               Benjamin Van Durme},
  title     = {Robsut Wrod Reocginiton via Semi-Character Recurrent Neural Network},
  booktitle = {Proceedings of the Thirty-First {AAAI} Conference on Artificial Intelligence,
               February 4-9, 2017, San Francisco, California, {USA.}},
  pages     = {3281--3287},
  year      = {2017},
  url       = {http://aaai.org/ocs/index.php/AAAI/AAAI17/paper/view/14332},
  publisher = {{AAAI} Press},
}

Data

.
├── README.md       # this file
├── train.py        # script for training
├── predict.py      # predict correct word
├── models          # store model files
├── binarize.py     # utility function
└── jumble_words.py # script for adding jumble noise

Basic Usage

(prerequisites)

    # Tested on CentOS7 (with CUDA7.5)
    > conda create -n robsut python=2.7
    > pip install keras theano h5py
    # change the keras backend to theano (edit $HOME/.keras/keras.json)
    

(training) 

    THEANO_FLAGS=device=gpu0,floatX=float32 python train.py

(predicting) 

    THEANO_FLAGS=device=gpu0,floatX=float32 python predict.py -m models/train_j-INT_n-JUMBLE_u-650_batch-20_ep-10_model.h5

Questions?

  • Please e-mail to Keisuke Sakaguchi (keisuke[at]cs.jhu.edu).