SVD + LDA Image Classification Project (MATLAB Starter)

This repository is a starter kit for the final programming project in MATLAB.

You will implement the following functions, each in its own .m file:

• power_method.m — dominant eigenvalue/eigenvector of a symmetric matrix.
• svd_compress.m — rank-k image approximation using SVD.
• svd_features.m — SVD-based feature extraction for images.
• lda_train.m — train a two-class Linear Discriminant Analysis (LDA) classifier.
• lda_predict.m — apply the trained LDA classifier.

The autograder will call these functions with specific signatures. Do not change the function names or input/output arguments.

Files

• power_method.m
• svd_compress.m
• svd_features.m
• lda_train.m
• lda_predict.m
• example_run.m — simple script that runs an end-to-end test using the example data.
• project_data_example.mat — tiny synthetic dataset for local testing.
• .gitignore — ignores editor/OS clutter and large data files.

Quick start (MATLAB)

1. Open MATLAB and add this folder to your path.

2. Open each of the .m files and implement the functions where indicated.

3. Run the script:

   example_run

   This will:

   • Load project_data_example.mat if present.
   • Extract SVD features using svd_features.
   • Train an LDA classifier with lda_train.
   • Predict labels on the test set with lda_predict.
   • Print the test accuracy on the tiny synthetic dataset.

In the real assignment, the Gradescope autograder will provide a different project_data.mat with more realistic images and labels.