KNN (K-nearest neighbours) classifier for the CIFAR-10 dataset. Apply 5-fold cross validation to find the best K and L (Euclidean or Manhattan distance) pair by plotting the resulting average accuracies and standard deviations and choosing the pair which produces the highest average accuracy first and lowest standard deviation second. Use this pair to find the accuracy per class and average accuracy on the testing set. Find out if grayscalse would produce a better accuracy than colored.
shungjhon / cifar-10-knn Goto Github PK
View Code? Open in Web Editor NEWThis project forked from therealyoussef/cifar-10-knn