Machine-Learning-Algorithms-Implementated From Scratch in Numpy

This Repo Contains Implementations of some of the fundamental Models in Machine Learning.

1. A Vectorized Numpy Implementation of a Decision Tree Learner Algorithm That works for n-nary features and Predictor Variables.

2. A Vectorized Numpy Implementation of Hidden markov models that trains by calculating the Emission, transition and Initial probabilities for given Sequence data and then uses a tabulized implementation of forward-backward Algorithm and viterbi algorithm for inference.

3. A Vectorized Numpy Implementation of Linear Approximation based Reinforcement Learning for obtaining the optimal policy for a car to reach a hilltop, given the car outright does not have enough energy to reach to the top. (Final Model utilizes both kinetic and potential energy to reach the top)

4. A Vectorized Numpy Implementation of a Natural Language Processing (NLP) system using binary logistic regression. Your algorithm will determine whether a restaurant review is positive or negative.

5. A Vectorized Numpy Implementation of a Nueral Network with various activation functions, loss functions, initialization schemes and Autodiff functionality of Pytorch using a subset of an OCR dataset.