202412121418
Status: #idea
Tags: Clustering Methods, Unsupervised Learning
State: #nascient
Model-Based Clustering
Instead of working based on density or graphs, but instead makes the logical assumption that data of each label were sampled out of some distribution that we try to estimate.
Since estimating some model is exceptionally difficult, we typically assume that each class is distributed according to some Gaussian distribution. This becomes the:
Gaussian Mixture Models