A variety of sample codes and experiments used to teach the course Math 649 (Foundations of Deep Learning) at Texas A&M University. This includes basic examples of function fitting using neural networks, image classification, generative modelling, physics-informed neural networks, and operator learning.
Contained in the folder basic-examples. Consists of one experiment fitting a sin function using a amultilayer perceptron (MLP) and gradient descent on the squared error loss.
Contained in the folder image-classification. Consists of two experiments, the first trains an MLP and the second trains a CNN to classify MNIST. The MLP training is done using vanilla gradient descent, while the CNN training is done using gradient descent with momentum.
Under construction
Under construction
Under construction