GitHub - Aashw1n/kaggle-wsdm-fake-news-classification: Model for kaggle fake news classification challenge

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 4 Commits
.ipynb_checkpoints		.ipynb_checkpoints
.gitattributes		.gitattributes
MaLSTM.ipynb		MaLSTM.ipynb
Report.pdf		Report.pdf
malstm.h5		malstm.h5
preprocessing.py		preprocessing.py
readme.txt		readme.txt
submission.csv		submission.csv
test.csv		test.csv
train.csv		train.csv
validataion.csv		validataion.csv

Repository files navigation

malstm.ipynb: ipython notebook made with Jupyter nootebook 
malstm.h5: HDF5 file where model is saved 

Dependencies: 
Keras 
NLTK 
Word2Vec 
Pandas 
Numpy 
Matplotlib

- Data was very noisy
- Not enough processing power to run for more epoch (Getting BSoD)
- Model was overfitted

- Done solo, this was kinda hard to get right. Need more theory and math.