COSC 40523 Deep Learning

Course Description

Introduces fundamental knowledge of deep learning algorithms, practical skills to build, train and tune a deep artificial neural network with applications such as computer vision, natural language processing, and business analytics. Topics include but not limited to fully connected neural networks, convolutional networks, recurrent networks, and generative adversarial networks.

Upon successful completion of this course, students will be able to

  • build, train, and apply fully connected deep neural networks for various classification problems
  • use the common neural network "tricks," including initialization, L2 and dropout regularization, gradient checking
  • implement and apply a variety of optimization algorithms, such as mini-batch gradient descent, Momentum, and check for their convergence
  • identify the key hyperparameters in a neural network's architecture
  • apply convolutional networks to visual detection and recognition tasks
  • apply sequence models to natural language problems, including text synthesis
  • build training pipelines and real-time inference run-times (TensorFlow and Python)
  • deploy and run deep neural models on GPUs

Prerequisites:College Calculus, Linear Algebra, Basic Probability and Statistics

Answering the following questions will tell you if you are ready to take the Machine Learning class. If you are not able to answer “Yes” to these questions, then we suggest that you go through the reading list at the end of this page.
  • Do you know matrix multiplication?
  • Do you know what are eigenvectors? Do you know how to calculate eigenvalues of a matrix?
  • Can you decompose a matrix using SVD (Singular Value Decomposition)?
  • Do you know what are the conditions of a valid distance metric?
  • Do you know what is the Bayes Rule?
  • Can you calculate the expectation of a random variable?
  • Can you calculate the covariance/correlation between two random variables?
  • Do you know gradient or slope?
  • Are you familiar with chain rule in calculus?
  • Have you worked with Python? All class assignments will be in Python.

Reading List

Free Online Stanford Machine Learning Course by Andrew Ng

  • Machine Learning by Professor Andrew Ng Click