Mathematics of Artificial Neural Networks

Catalog description

Prerequisites: MATH 1320 , MATH 2000 , MATH 2450 , and graduate standing. This course provides an introduction to the mathematical ideas and techniques underlying the modern theory of artificial neural networks. Guidance and training for implementing practical applications are also provided. Topics may include fundamentals of supervised learning, testing, and validation for parametric statistical models, feedforward neural networks, forward propagation, activation functions, loss functions, batch/stochastic/mini-batch gradient descent, the backpropagation algorithm, preventing overfitting, basics of convolutional neural networks (CNNs), other specialized architectures. Topics are identical to MATH 4750 , but the material is covered at a greater depth, and additional projects/assignments are required. Credit cannot be earned for both MATH 4750 and MATH 5750 .