[ 875BIMLDLNK16 ] KV Deep Learning and Neural Networks

Workload Education level Study areas Responsible person Hours per week Coordinating university
3 ECTS M1 - Master's programme 1. year (*)Bioinformatik Sepp Hochreiter 2 hpw Johannes Kepler University Linz
Detailed information
Original study plan Master's programme Bioinformatics 2016W
Objectives In recent years, Deep Learning has emerged as one of the most promising Machine Learning techniques. It is by far the best known method for image recognition, has has increased the performance in speech recognition by a large margin and powers the current Android phone voice recognition, it has been used to automatically translate from one language to another, to drive autonomous cars, to predict the biological activity of small chemical compounds, to automatically learn an AI for video games and much more.

This class explains the methods behind Deep Learning and its applications. Both mathematical details as well as practical aspects are shown. Students will implement the algorithms for performing classification and regression tasks, and try it out on applications such as image classification or audio compression. After attending the class, students should have an thorough understanding of how deep neural nets work, should be able to understand the current research literature on the topic and apply Deep Learning to new problems.

Basic knowledge of machine learning is expected, as well as familiarity with basic linear algebra, calculus and statistics. All programming examples and homework problems will be in Python.

Subject Short Recap: Supervised Machine Learning, Neural Networks and the backpropagation algorithm, Auto-Encoders and Restricted Bolzman Machines (RBMs), Deep Neural Networks and Dropout, Convolutional Neural Networks, Recurrent Neural Networks and LSTMs, Advanced Algorithms and Applications
Criteria for evaluation The final grade is based on a combined assessment of homework and a final exam.
Language English
Changing subject? No
On-site course
Maximum number of participants 35
Assignment procedure Direct assignment