Neural Networks and Deep Learning with Object Oriented Programming

Week I - Week IV

Task 1: Create a grade A symbol using a neural network.

Task 2: Create a Multi-Layer Perceptron Object, trained and optimized the algorithm using the Gradient Descent.

Week V - Week VI

Task 1: Use NNs to classify wine dataset. Used 4 different approaches for the same, first is a multi-layer perceptron without normalization, second uses linear normalization, third uses z-score normalization and finally used one hot encoding with categorical cross entropy loss function with adam optimizer.

Week VII - Week X

Task 1: Use Convolutional Neural Network to classify MNIST digit dataset. Achieved 99.12% accuracy with CNN

Task 2: Comparison of MLP and CNN architectures to classify the Fashion MNIST Dataset. Achieved Max test accuracy of 89.76% for MLP and 93.43% for CNN.