Special Course in Machine Learning: Unsupervised Neural Networks with TensorFlow
MTAT.03.317
Seminars: Thursdays 14:15, at Paabel (Ülikooli 17), room 219
Contact:
Kairit Sirts, kairit.sirts@ut.ee, room 218
Overview
This course is about unsupervised learning with neural networks and how to implement them using the TensorFlow library. In particular, we will discuss various autoencoders and the restricted Boltzmann machines. Depending on the preliminary knowledge of the participants, we might also cover variational autoencoders and generative adversarial networks. We will start with a short introduction to artificial neural networks and training with back-propagation, which should make the course accessible also to people who are not super familiar with neural network learning.
Process
This course has the seminar format, which means that there will be home readings/practical work assigned for each week and the seminar time is used to discuss and clarify the material. Most seminars will also include a test about the reading materials. Topics will be divided among students and the person responsible for the topic will prepare and later also grade the test.
Materials
The course is mainly based on two resources:
- Deep Learning book
- Deep Learning in TensorFlow course in Big Data University
Grading
This is a pass/fail course. In order to pass the course each student must:
- Attend the seminars and participate in discussions
- Read the assigned materials
- Do all practical homework
Preliminaries
In order to succeed in this course the students should be familiar with:
- Linear algebra and probabilities
- Machine learning in general
- Programming in python