DSBA 6165: Deep Learning & Artificial Intelligence

Course Overview & Objectives

This course aims to introduce state-of-the art methods in Deep Learning and build understanding of its sub domains through hands-on programming and projects. Broadly course is divided into following topics:

• Neural Networks: Building neural networks for classification and regression, understanding loss functions, optimization methods like gradient descent, initialization strategies, the importance of depth etc.
• Computer Vision: Building Convolutional Neural Networks, State-of-the art architectures, Image Classification, Object Detection, Instance Segmentation and other computer vision applications.
• Natural Language Processing: Recurrent Networks, Embeddings, Transformers, Language Models.
• Generative Modeling: GANs, VAEs, Stable Diffusion, GenAI

Additionally, the following topics are tentative and may be added to the schedule depending upon the pace:

• Graph Neural Networks, Reinforcement Learning, Introduction to LLMs, and AI Ethics.

Course Material

All material will be distributed through Canvas.

Lecture Slides

• Lecture notes are provided in the form of slides and will contain references to all the material needed for each topic.
• Numerous research papers, articles and topics from the books are referenced and appropriately cited where discussed.

Reference Books

• Dive into Deep Learning by Aston Zhang, Zachary C. Lipton, Mu Li, and Alexander J. Smola
• Deep Learning by Ian Goodfellow, Yoshua Bengio, and Aaron Courville
• Understanding Deep Learning by Simon J.D. Prince

Videos

• To further aid in understanding, students are encouraged to watch additional 2-5 minute videos on certain topics as necessary.

Course Format

• Quiz: Based on the content covered in the lectures and the videos posted on canvas, we will have a short poll everywhere quiz in each class.
• Labs: In the latter half of each lecture, students will get to work on hands-on programming labs (Jupyter notebooks) which will be based on the topic covered in the lecture that day.
• Assignments: We will have 2 individual programming assignments.
• Group Project: Students will engage in a semester long group project working on a real-world deep learning problem. The project will be divided into 6 stages, each targeting on finishing a milestone. Details about the project available on Canvas.

Grading

Prerequisites

• Knowing how to program in Python
• Some understanding of machine learning and familiarity with data manipulation libraries like NumPy and Pandas

Tools / Skills to develop

• PyTorch, TensorFlow