Supervised Learning Project 2

This repo contains the code for the implementations of various classifiers on the popular MNIST dataset. source_code_p1.ipynb contains the code for implementing a kernel perceptron and running multiple experiments on the dataset. source_code_p2.ipynb contains the manual implementations for 4 basic algorithms: the perceptron, the winnow, least squares regression and K-nearest Neighbors. Lastly, p1_svm_manual_implementation.ipynb contains a manual implementation of the support vector classifier, which is connected to the task in source_code_p1.ipynb.

The results for all my models are accessible in the folders part_1_results/ and part_2_results/.

For a detailed discussion of my methods and results, checkout my report in report.pdf.

Requirements for running the code

• NumPy
• CuPy
• SciPy
• MatPlotLib
• Seaborn
• Pandas
• CVXOPT
• itertools, os, time