Dimentionality Reduction

Project information

Dimensionality Reduction and Classification Methodology

Load the MNIST dataset of handwritten digits Perform the dimensionality reduction methods that were introduced in the lecture: (a) PCA (b) IsoMap (c) Local Linear Embedding (LLE) (d) t-distributed stochastic neighbor embedding (t-SNE) (e) Uniform maniforld approximation and projection (UMAP) Reduce the dimensionality to d = 2 dimensions and plot the results Use the low dimensional representations to perform unsupervised clustering with k-means. Use k =10 classes for the clustering algorithm Classify all images according to the clustering. Choose the label based on how many the majority of datapoints within a cluster

dataset Python