Second Assignment in 'Practical topics in Machine Learning' course by Dr. Kfir Bar at Bar-Ilan University.
In this assignment, I had to build a convolutional network (CNN) and train it over the SVHN data, which is a collection of 3X32X32 images, classified into 10 digits.
I trained the model once on CPU and once on GPU and compared the training time and the accuracy on both training set and test set.
Score: 100