Recognizing MNIST Handwritten Data Set Using PCA and LDA

Abstract

In data analyzing and machine learning, PCA and LDA are widely used tools, these are linear transformation methods to reduce the dimension observation, and these are used in many practical applications like compression and data visualization in machine learning. Here, I have applied these two methods that are principal component analysis (PCA) and linear discriminant analysis (LDA) in handwritten digit recognition. The main motto of this paper is to present a simple and clear understanding of these methods. Here, this paper depicts that LDA can outperform PCA when training data set is huge, and PCA can outperform LDA when training data set is small.