The Robustness of SVM Kernels to Noise: A Comparative Analysis on MNIST

About this project

A study testing SVM kernel robustness (Linear, Polynomial, RBF) against noisy MNIST data. While RBF leads on clean data at 96.50%, it degrades fastest under stress. The Polynomial kernel proved most robust, maintaining nearly 40% accuracy even at 100% noise. This project explores the bias-variance tradeoff and why highly flexible models can struggle with messy, real-world inputs.