IT-3030: Deep Learning

This is the instructor's page for IT-3030. We use this web page for (relatively) static content, while online meetings, discussions, etc. are found on Blackboard.

This course covers the theoretical basis of deep neural networks, including both the calculus and linear algebra underlying deep learning's cornerstone algorithm: backpropagation. In addition, students get hands-on experience in both a) programming deep networks from scratch, and b) using popular deep-learning software (e.g., Tensorflow, PyTorch, etc.) for complex classification tasks.

Key concepts covered in this course include: standard backpropagation nets, convolution nets, recurrent nets (e.g. long short-term memory, LSTM), regularization techniques, optimizers, autoencoders, generative adversarial networks (GANs), representation learning, structured probablistic models (e.g. variational autoencoders, VAE), transformers, and more. See the Lecture plan for more details.

People

Instructors

• Boye Høverstad (primary contact person)
• Keith Downing

PhD Assistant

• Bjørnar Vassøy

Teaching Assistants

• Katja Sivertsen
• Olav Førland

When and Where

Lectures

• Where: Room GL-GE G1
• When: Tuesdays 14:15 - 16:00
• First meeting: January 9, 2024

Teaching Assistance and Demos

TA hours and location are subject to change.

• TAs available Thursdays 10:15 - 12:00 in GL-KH KJL24.
• Demo locations will be decided on a per-event basis. Students will need to register for a time slot to give demos. More information will be given on Blackboard.

Exam

• When: May 29, 2024 @ 15:00
• More info under Exam and on the official web page for the course.

Important Links

NTNU's official web page for IT-3030.