Classifying-Hand-Written-Digits---Support-Vector-Machines

A Machine Learning Project that aims to classify handwritten digits from 0 to 9. Support Vector Machine Algorithm is used to solve this challenge. The training set (mnist_train.txt) contains 2000 digits, and the test set (mnist_test.txt) contains 1000 digits. Each line represents an image of size 28×28 by a vector of length 784, with each feature specifying a grayscale pixel value. The first column contains the labels of the digits, 0–9, the next 28 columns represent the first row of the image, and so on. Gaussian Kernel is applied on Multiclass non-linear SVM to classify numbers in 10 classes (0-9). Different values of C and gamma parameters are used and then cross-validated to get the lowest error-percentage. The code takes some time to run because of the cross validation (5 folds) on 8 different values of C and gamma.