MA 416 - Deep Learning
- Credit Hours: 4R-0L-4C
- Term Available: Arranged
- Graduate Studies Eligible: No
- Prerequisites: MA 212 or MA 221, and either MA 223 or MA 381, and either CHE 310 or CSSE 220 or ECE 230 or MA 332 or MA 386 or ME 323 or ME 327 Prerequisite Clarification for MA416: Junior standing and either MA212 or MA221, and either MA223 or MA381, and one of CHE310, CSSE220, ECE230, MA332, MA386 or (ME323 or ME327).
- Corequisites: None
An introduction to deep learning using both fully-connected and convolutional neural networks. Topics include: least squares estimation and mean square error, maximum likelihood estimation and cross-entropy, convexity, gradient descent and stochastic gradient descent algorithms, multivariate chain rule and gradient computation using back propagation, linear vs nonlinear operations, convolution, over-fitting vs under-fitting and hyper-parameter optimization, L2, early stopping and dropout regularization, data augmentation and transfer learning.