1- Implementation of PCA from scratch.
a) implementation of covariance matrix.
b) implementation of eigenvalues and eigenvectors using power iteration method.
2- Preprocessing the mnist dataset making it binary image (only zeros and ones) to apply the hamming network later on it.
3- Tring different number of components in PCA till gets best result.
4- Cluster data using k-means (you can replace it with any clustering technique).
5- Apply Hamming on unseen data point with PCA and without PCA.