Understanding Machine Learning through Animated Visualizations
This repository contains the source code for animated visualizations of some famous machine learning algorithms. They were created using the R package animation, and ilustrate algorithm convergence and the effect of hyper-parameter tuning. The animations available so far are:
- XGBoost decision boundary as iterations proceed:
- KNN decision boundary varying the number of nearest neighbors k.
- Multivariate Gaussian Mixture Models (GMMs) fitting by EM algorithm.
- Multimodal Density Estimation using GMMs.
- Tikhonov Regularization effect in Extreme Learning Machines (ELMs).
- Image Segmentation using K-means.
- Image Reconstruction using Principal Components Analysis (PCA).
Take a look at the website and have fun!