we will explore
mathematics that
underlies modern machine learning techniques. We will
dive into a
selection of topics such as
- Probably approximately correct learning
- Self-organizing neural networks such as Hopfield
networks and
Kohonen maps,
- Math behind the famous backpropagation
- Universal approximation theorem
- Independent component analysis
The students will be able to influence where we focus
on, where we go
fast and which topics we select. The focus of the course
is theoretical,
although we might have some practical programming
exercises.
Background in mathematics is useful but not absolutely
necessary. If you
have done any university level course in mathematics
such as linear
algebra or functional analysis, it will make the course
easier for you.
Only interest in mathematics is necessary ;)
- Opettaja: Nicoletta Prencipe
- Opettaja: Vadim Weinstein