1 - Performed linear regression of a noisy sinewave using a set of gaussian basis functions with learned location and scale parameters. Model parameters were learned with stochastic gradient descent.
2 - Solved the Two Spirals Problem by performing binary classification on the spirals dataset, consisting of 3 turns, with a neural network. Network consisted of 1 hidden layer with 16 neurons, with sine used as the activation function.
3 - Classified MNIST digits with a convolutional neural network. Attained 97.74% accuracy on the test set.
4 - Classified CIFAR10 and CIFAR100 with a convolutional neural network. Applied data preprocessing and data augmentation. Attained 81.74% accuracy on the CIFAR10 test set and 50.64% accuracy on the CIFAR100 test set.