- Be able to design your own machine learning algorithms tailored to the problems you face in research
- Have a better grasp of which models and algorithms to use in which situations
- Understand basic mathematical theory underlying machine learning methods
- We do not focus on implementation, but on theory.
This course can replace other (compulsory or not) machine learning courses in your degree programme. Please check with your programme supervisor whether this is the case.
Topics covered:
Topics covered:
- Neural networks, perceptrons, McCulloch-Pitts units
- Backpropagation, chain rule, computational graphs
- Universal approximation theorems
- Hopfield networks (the subject of 2024 Nobel prize)
- Independent component analysis
- Reinforcement learning
- Generative models (models behind LLMs)
Prerequisites
- Basic understanding of derivatives, matrices, and probabilities (we will have review sessions)
Schedule
- Lectures Mon & Fri 12-14
- Exercise Wed 16-18
- First half of the course is lectured live, second half is lectured online.
- First week exceptions: The second lecture is delivered on Wednesday 30.10 at 16-18 instead of the exercise class, and there is no lecture on Friday 1.11. There is no exercise class on the first week.
- First lecture October 28th 2024 at 12:15 in L1
Course details:
- Peppi/Moodle ID: 521243S-3004
- Period 2
- Official name: "Special Course in Information Technology 11 - Mathematical foundations of methods in machine learning"
- 5 ECTS
- Department of electrical engineering
- Lectured by Nicoletta Prencipe and Vadim Weinstein
- Exercise class by Filip Georgiev
- Contact us at vadim.weinstein@oulu.fi, nicoletta.prencipe@oulu.fi, or filip.georgiev@oulu.fi
- Lärare: Filip Georgiev
- Lärare: Nicoletta Prencipe
- Lärare: Vadim Weinstein