Skip to content

Fake news classification with Keras: Bi-GRU with SpaCy and custom embeddings

License

Notifications You must be signed in to change notification settings

louismagowan/fake_news_classifier

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Contributors Forks Stargazers Issues LinkedIn

Fake News Classification with Keras

About The Project

This project looks at building, training and evaluating neural networks to solve the fake news classification problem on Kaggle. It uses Python and Keras.

Gated Recurrent Units (GRUs) are used, along with batch normalization and both spaCy and Keras embedding layers.

Keras embedded models appear to outperform those with a static spaCy embedding and manage to achieve extremely high test accuracies (99%+).

The run_models.ipynb notebook contains a demo of how code from the two Python files can be used to easily implement the networks needed to solve this task.

Caveat: The Kaggle dataset is not ideal- there are leakages (that I've tried to remove) and issues around its provenance. However, the papers that produced the dataset are reasonably well cited (~200 citations) on Google Scholar. In any case, this project is presented as more of a fun tutorial/look at how deep learning approaches could be used to solve the problem of fake news identification, rather than any serious, real-world classifier.

Feel free to check out the Medium article on this here

(back to top)

Built With

(back to top)

Contact

(back to top)

Acknowledgments

(back to top)

About

Fake news classification with Keras: Bi-GRU with SpaCy and custom embeddings

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published