https://github.com/chen0040/keras-fake-news-generator-and-detector
Fake news generator and detector using keras
https://github.com/chen0040/keras-fake-news-generator-and-detector
fake-news-classification glove-embeddings keras lstm-neural-networks
Last synced: 2 months ago
JSON representation
Fake news generator and detector using keras
- Host: GitHub
- URL: https://github.com/chen0040/keras-fake-news-generator-and-detector
- Owner: chen0040
- License: mit
- Created: 2017-12-24T13:59:51.000Z (almost 8 years ago)
- Default Branch: master
- Last Pushed: 2018-02-14T02:38:39.000Z (over 7 years ago)
- Last Synced: 2025-08-15T22:50:08.591Z (2 months ago)
- Topics: fake-news-classification, glove-embeddings, keras, lstm-neural-networks
- Language: Python
- Size: 36.2 MB
- Stars: 24
- Watchers: 3
- Forks: 7
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# keras-fake-news-generator-and-detector
Fake news generator and detector using keras
The fake news data is from [https://github.com/GeorgeMcIntire/fake_real_news_dataset](https://github.com/GeorgeMcIntire/fake_real_news_dataset)
The deep learning models are implemented in the [keras_fake_news_detector/library](keras_fake_news_detector/library) folder
# Fake News Detector
The demo codes for neural network models implemented for fake new detector can be found in [demo](demo) and are listed bel:
* LSTM recurrent network with embedding layer
* training: run [demo/lstm_classifier_train.py](demo/lstm_classifier_train.py) to train and save the model
* predicting: run [demo/lstm_classifier_predict.py](demo/lstm_classifier_predict.py) to load the trained model for prediction
* Feed-forward network with GloVe embedding layer
* training: run [demo/glove_ffn_classifier_train.py](demo/glove_ffn_classifier_train.py) to train and save the model
* predicting: run [demo/glove_ffn_classifier_predict.py](demo/glove_ffn_classifier_predict.py) to load the trained model for prediction
* Feed-forward network with Doc2Vec that encode the new article and pass as input to the feedforward network
* training: run [demo/glove_ffn_classifier_train.py](demo/glove_ffn_classifier_train.py) to train and save the model
* predicting: run [demo/glove_ffn_classifier_predict.py](demo/glove_ffn_classifier_predict.py) to load the trained model for prediction
Below are the training history in terms of loss and accuracy for a number of neural network implemented in keras:
