Implements the perceptron, average perceptron and pegasos from zero using numpy. Provides some data processing functionality. A model is trained and refined for sentiment analysis of a sample of Amazon product reviews.
Mnist folder contains several implementation of algorithms to classify mnist digits. Includes softmax regression, svm and kernel svm, ANN implementations with numpy, CNN using pytorch.
Mixture folder contains implementation of Gaussian mixture model which is used to predict movie ratings.