This will be a 16 hour lecture course on the theory and applications of deep learning. All lectures should be recorded automatically and available shortly afterwards through the class moodle. The videos are also directly available via Panopto.
Live lectures will be recorded, and we assume that if you have your camera or microphone on, then you consent to being recorded. If you wish to revoke your consent, please contact us and we will need to edit you out of the recording.
[2023-10-26 Thu 09:00-11:00]
[2023-10-30 Mon 09:00-11:00]
[2023-11-02 Thu 09:00-11:00]
Back propagation continued; derivation.
[2022-11-07 Mon 09:00-11:00]
[2023-11-09 Thu 09:00-11:00]
Autograd
[2022-11-14 Mon 09:00-11:00]
Images ctd.
Sequences (part 1)
Sequences (part 2)
Transformers
[2023-11-16 Thu 09:00-11:00]
https://arxiv.org/abs/2304.10557
RL [2023-11-20 Mon 09:00-11:00]
This optional session will be 1-3pm on Friday 10th November in MR15.
Most of the lectures will mention key papers to read. A good introductory text is AI Engines: mathematics of deep learning by Dr James V Stone.
One assignment to be set at the end of the course; due in start of Lent Term 2024.
See https://github.com/sje30/dl2023/wiki for questions and comments on the assignment. If you have a question ahead of the feedback session on 7th December (10am, zoom), please email me.