1. Your task is to develop a code for PCA (hence construct eigen-digits) to project input digit images into eigen-digit space and find the matching digit index (0 to 9) given a test input digit image.
In this project, you will develop a handwritten digit (from MNIST dataset) recognition code. Your system (hence a program code) reduces each digit image to a vector, then uses Principal Component Analysis (PCA) to find a linear subspace for 10 digits.
Upload, and show any sample images in Matlab or Python (MNIST dataset).
The input data you will be using is divided into several sets. They can be downloaded from this link. http://yann.lecun.com/exdb/mnist/
2. PCA is a technique by which we reduce the dimensionality of data points
Describe the steps in the PCA for face recognition (limit to half page length)
Please see the Lecture notes for details.
3. Write a pseudo-code for 2 (limit to half page length)
4. Using the results from problem 1 to 3, construct a code that distinguishes 10 classes from 0 to 9 from a given dataset (MNIST) using PCA, so-called hand-writing recognition.
In your code, show how accurate the test image in terms of classification performance is. Also show how a set of k coefficients is obtained by a code implementation.
Helpful links:
https://courses.cs.washington.edu/courses/cse455/09wi/projects/project4/web/project4.html
https://static1.squarespace.com/static/55133727e4b05ad39f9b9749/t/56ba9b20a3360ce1d09cb271/1455069985248/principal-component-analysis.pdf
https://medium.com/analytics-vidhya/principal-component-analysis-pca-with-code-on-mnist-dataset-da7de0d07c22
