MATLAB Machine Learning and Image Processing Project

Author: Nichita-Adrian Bunu, 323CA Facultatea de Automatica si Calculatoare UNSTPB

Contact: nichita_adrian.bunu@stud.acs.upb.ro

---

Overview

This project contains a set of MATLAB scripts designed for machine learning and image processing tasks. The primary components focus on implementing K-Nearest Neighbors (KNN) for classification, preparing data for analysis, visualizing images, and utilizing Principal Component Analysis (PCA) for dimensionality reduction.

---

Features and Functionality

The main features include:

• K-Nearest Neighbors (KNN) Classification: A basic implementation for predicting labels based on the nearest neighbors.
• Data Preparation: Functions for preprocessing and organizing data for machine learning tasks.
• Principal Component Analysis (PCA): Used to reduce the dimensionality of data for more efficient processing.
• Image Visualization: Tools to reconstruct and display images from datasets.

---

Scripts Overview

KNN.m

Implements the K-Nearest Neighbors algorithm:

• Calculates the Euclidean distance between the input test sample and each training sample.
• Sorts distances to find the nearest neighbors and determines the label by taking the median of the nearest labels.

classifyImage.m

Performs classification of an image based on trained data using the KNN algorithm:

• Loads the dataset and prepares it for classification.
• Uses KNN to classify the provided image data.

magic_with_pca.m

Handles dimensionality reduction using PCA:

• Computes the principal components from the data matrix.
• Projects data onto a reduced dimensional space for better efficiency in processing and visualization.

prepare_data.m

Prepares datasets by normalizing and structuring data for analysis:

• Centralizes data by removing the mean.
• Adjusts data to improve the performance of machine learning algorithms.

prepare_photo.m

Specifically prepares image data for PCA processing:

• Reshapes and normalizes image matrices.
• Handles conversion between different data formats.

task1.m, task2.m, and task3.m

Contain various tasks for performing specific machine learning experiments and visualizations:

• May include tasks like feature extraction, training different models, or applying transformations.

visualise_image.m

Reconstructs and visualizes images from data matrices:

• Reshapes data back into the original image format for display.
• Converts processed data into a visual format suitable for interpretation.

---