Data from https://www.kaggle.com/competitions/digit-recognizer/data PCA implement by 1- calculating covariance matrix 2- using sklearn tSNE using sklearn
Variational Auto-Encoder on MNIST data using pytorch. Many examples and explanation on this but I could not find a reference that explains how the loss function is created. This lecture is very useful Yotube but there also the derivation of the loss function is not done but general concept is explained in detail. The same data as above is used.
Three application of K-means are tried.
- It is used on Olivetti faces dataset for classification Link.
- It is used as a non-linear dimension reduction
- It is used to label a dataset.
The notebook shows some comments and discussed the limitations.