Natural-Language-Processing

1. Spam SMS detection using Naive Bayes
   • Cleaning text data.
   • Naive Bayes implementation from scratch.
   • Naive Bayes implementation using scikit-learn.
2. Topic Modeling - LDA
3. Sentiment analysis with regular classifier - GradientBoostingClassifier, GaussianNB
   • Test accuracy - 72%
4. Sentiment analysis with LSTM
   • Test accuracy - 86.5%
5. Attention basics
   • https://arxiv.org/pdf/1706.03762.pdf
6. Deciphering Code with Character-Level RNN
   • Pre-process text to convert encoded text to matrix of integers
   • Used GRU RNN
   • output - Vector of decoded text