bala-codes/Fake_Vs_Real_News_Classifier_Using_Machine_and_Deep_Learning

In this repo, I've built an Machine and Deep Learning-based classifiers (Ensemble) to distinguish Real Vs Fake news from the news data.

FAKE VS REAL NEWS CLASSIFIER USING MACHINE AND DEEP LEARNING.

In this repo, I've built an Machine and deep learning based classifiers (Ensemble) to distinguish Real Vs Fake news from the news data.

MACHINE LEARNING

Machine Learning algorithms (sklearn) used in this projects,

• LogisticRegression
• LogisticRegressionCV
• MultinomialNB
• BernoulliNB
• SVC
• LinearSVC
• PassiveAggressiveClassifier
• SGDClassifier
• CalibratedClassifierCV
• KNeighborsClassifier

After Hyperparameters tuning, the best Machine Learning Ensemble Score is 90%

Checkout the ML based classifiers part 1 at and part 2 at

VANILLA NEURAL NETWORK, CNN + WORD EMBEDDINGS

• The Deep Learning (NN) Scores are : Training Accuracy: 0.9871 & Testing Accuracy: 0.8953

Checkout at

• The Deep Learning (CNN + W_Embedding) Scores are : Training Accuracy: 0.9849 & Testing Accuracy: 0.8763

Checkout at

LSTM AND WORD EMBEDDINGS

• The Deep Learning (LSTM + Keras Embedding) Scores are : Training Accuracy: 0.9998 & Testing Accuracy: 0.9002

Checkout at

• The Deep Learning (LSTM + Glove Embedding) Scores are : Training Accuracy: 0.9723 & Testing Accuracy: 0.9077

Checkout at