COMP 551 Mini Project 4

2019-04-17 Segev, Michael Jacquier, Pierre Han, Zhenze

Models and experiments are split in seperate python scripts that all use common classes to load files and save models to file.

1. NaiveBayesBench.py Run this file to test different feature extraction pipelines with a NB classifier on Stanford Sentiment Treebank.

2. SVMBench.py Run this file to train and test/validate Support Vector Machine model on Stanford Sentiment Treebank.

3. RNNBench.py Run this file to train and test/validate Recursive Neural Network model on Stanford Sentiment Treebank.

4. DecisionTreesBench.py Run this file to train and test/validate extremely random trees model on Stanford Sentiment Treebank.

5. metaClassifier.py Run this file to train and test/validate stacking ensemble meta-classifier on on Stanford Sentiment Treebank using pre-trained models saved as pickle files.

Original Paper

@inproceedings{le2014distributed,
	title={Distributed representations of sentences and documents},
	author={Le, Quoc and Mikolov, Tomas},
	booktitle={International conference on machine learning},
	pages={1188--1196},
	year={2014}
}

Library Dependecies:

re
numpy
scikit-learn
keras
tensorflow-gpu

Copyright

MIT license