For each category the top 20 eigenvectors of the data covariance matrix are calculated. The images are represented using these eigenvectors and projected into a reduced space. The error resulting from representing images of each category using these vectors is calculated and plotted.

The similarity of 2 classes can be computed by calculating the distance between their mean images. More distance corresponds to lesser similarity and vice versa. Distance function used : 2-norm. Here is the matrix for the similarity scores of the 10 classes:

Another measure of similarity between two classes A and B can be defined as : sim(A,B) = 1/2( E(A->B)+E(B->A))) where,E(A->B) is defined as the average error obtained by representing all images of class A using the mean of class A and the first 20 principal components of class B. This is the list of the 3 most similar classes for each class in the dataset obtained from this similarity measure.