IIITH NLP Lab Summer Research Project

Beyond Word2Vec

• Vijay Prakash Dwivedi
• Dr Manish Shrivastava

Description

Building deep learning architectures to embed multi-word units into a vector space maximizing similarity between units of different sizes.

Work

Implemented a Siamese MLP architecture with following best results.

* Training on 263000 samples, Testing on 113000 samples
* Accuracy on training set: 91.06%
* Accuracy on test set: 74.93%

Also, implemented a Siamese LSTM architecture with following best results.

* Training on 263000 samples, Testing on 113000 samples
* Accuracy on training set: 93.20%
* Accuracy on test set: 76.65%