ML-from-Scratch contains my personal implementations of Machine Learning algorithms, built with Python's Numpy, Implementing everything from scratch was a great way for me to understand how the algorithms work and I hope it'll be the case for you too.
This repository contains:
- Mini_Bactch/Batch & Stochastic gradient descent linear Regressor
- Mini_Bactch/Batch & Stochastic GD logistic (cross-entropy) Regressor
- KMeans Clustering Algorithm
- Multivariate Gaussian Distribution for Anomaly Detection
- Principal Component Alalysis (PCA) for Dimension Reduction
- Fully Connected Dense Neural Networks
I don't recommend using the implementations for any other reason than learning, highly opimized Machine Learning librairies such as Tensorflow (Keras) and Scikit-Learn do the job faster and often offer GPU support.
All you need to run the files is python's numpy