/Image-Analysis-with-PCA

Machine learning homework exploring image analysis and PCA dimensionality reduction.

Primary LanguageJupyter NotebookMIT LicenseMIT

Machine Learning Image Processing and PCA

This repository contains the code and documentation for a machine learning homework assignment. The assignment involves tasks related to image preprocessing, dimensionality reduction using PCA (Principal Component Analysis), classification using k-Nearest Neighbors (k-NN), and Non-negative Matrix Factorization (NMF).

Introduction

This machine learning homework focuses on the following tasks:

  1. Image Preprocessing and Visualization: Load a set of images, resize them to a common dimension, reshape them into 1D arrays, and apply PCA for dimensionality reduction. Visualize the images in a 2D space.

  2. Explanation of PCA Results: Provide an explanation of the PCA results and the significance of image positions in the 2D space.

  3. Classification Using k-Nearest Neighbors (k-NN): Perform image classification using the k-NN algorithm in both the original image space and the reduced PCA space. Evaluate the classification accuracy using 5-fold cross-validation.

  4. Optimal Number of Principal Components for PCA: Experiment with different numbers of principal components in PCA and identify the optimal number that yields the best results.

  5. Non-negative Matrix Factorization (NMF): Implement NMF with regularization on a synthetic dataset. Explore different parameter settings and track the number of iterations required for convergence.

Prerequisites

Before running the code in this repository, make sure you have the following prerequisites:

  • Python (version specified in the code)
  • Required Python libraries (e.g., scikit-learn, NumPy, OpenCV, PIL)
  • Jupyter Notebook or an integrated development environment (IDE) for running Python scripts

Usage

  1. Clone this repository to your local machine:
git clone https://github.com/NikosMav/Image-Analysis-with-PCA.git

  1. Open the Jupyter Notebook or Python script provided in the repository.

  2. Follow the instructions in the notebook/script to run the code for each task.

  3. Examine the results and documentation provided in the notebook/script to understand the outcomes of each task.

Results

  • The results of the PCA visualization, k-NN classification, and NMF experiments can be found within the provided Jupyter Notebook or Python script.

License

This project is licensed under the MIT License - see the LICENSE.md file for details.