drawing caption
Natural Scene Image
Handwritten digit recognition is a fundamental and widely studied task in the fields of computer vision and pattern recognition, serving as a benchmark for evaluating machine learning models. In this project, we conducted an extensive comparative analysis of multiple classical machine learning algorithms on the MNIST dataset, which comprises grayscale images of handwritten digits (0–9). Our objective was to understand and evaluate the individual strengths, limitations, and trade-offs of various models — namely K-Nearest Neighbors (KNN), Decision Trees, Logistic Regression, Random Forests, Naive Bayes, and Support Vector Machines (SVM) — with respect to both classification accuracy and computational efficiency. Each algorithm was trained using 60,000 labeled training images and evaluated on a separate set of 10,000 test images. To provide a detailed assessment beyond simple accuracy, we computed class-wise evaluation metrics such as precision, recall, and F1-score. These helped capture performance nuances for specific digits, particularly in cases where visual ambiguity made classification harder. Our experimental results reveal how different models behave under the same preprocessing and data distribution, and they shed light on the practical applicability of these techniques in real-world image classification scenarios. This project not only reinforced our understanding of fundamental ML concepts but also illustrated the importance of model selection based on dataset characteristics and task requirements.
Keywords: Handwritten Digit Recognition, MNIST Dataset, Machine Learning, Classification, K-Nearest Neighbors (KNN), Decision Trees, Logistic Regression, Random Forests, Naive Bayes, Support Vector Machines (SVM), Precision, Recall, F1-Score, Model Comparison, Image Classification
Bibtex
@Misc{PRML2025Course,
author = {Nitesh,Nitin Verma,Saurabh Kumar, Vanshita Jeenwal,Shruti Sunil Vibhute,Mallam Vishnu Priya},
title = {Handwritten Digit Recognition Using Classical Machine Learning Algorithms},
howpublished = {Course Project, Pattern Recognition and Machine Learning, Indian Institute of Technology Jodhpur},
year = {2025},
note = {Unpublished undergraduate course project},
}