Projects roughly from Udacity NLP nanodegree. Below them are some topics I've covered so far.
- Sentiment Analysis in 30 lines (details)
- Caesar Cipher Decryption (details)
- English to French machine translation (details)
Language Structure - part of speech tagging, hidden Markov model (HMM), Bayes theorem, Bag of Words
Math for Topic Modelling - GLoVe, Word2vec, Latent Dirichlet allocation (LDA), Dirichlet distributions (alpha-beta), matrix factorization
RNN Networks - simple RNN, Bidirectional RNN, Sequence to Sequence, LSTM, Additive Attention -> Multiplicative Attention -> Self-Attention
Traditional Automatic Speech Recognition (ASR) - voice user interface, fast fourier transform (library implementation), Mel Frequency Cepstral Coefficients MFCCs), phonetics, HMM implementation, N-grams Maximum Likelihood Estimation (MLE)
New Paradigm for ASR - Deep learning, CNN for audio feature extraction, RNN for time series data, CTC (Connectionist Temporal Classification) for audio-word sequencing