/Named-Entity-Recognition-for-Indian-Culinary-Science

Primary LanguageJupyter NotebookMIT LicenseMIT

Residual Stack LSTM with Biased Decoding

The implementation differs from the original paper in these ways:

  1. no lexicons
  2. Nadam optimizer used instead of SGD
  3. Parameters: LSTM cell size of 200 (vs 275), dropout of 0.5 (vs 0.68)

Result

The implementation achieves a test F1 score of ~86 with 30 epochs. Increase the number of epochs to 80 reach an F1 over 90.

Dataset

Indian Culinary Dataset GloVe vector representation from Jeffrey Pennington, Richard Socher, and Christopher D. Manning. 2014. See https://nlp.stanford.edu/projects/glove/

Dependencies

1) numpy 
2) Keras
3) Tensorflow
4) Stanford GloVE embeddings