Courses – Fakultet tehničkih nauka u Novom Sadu

Subject: Deep learning methods (17.IZMI03)

Type of studies	Title
Master Academic Studies	Mathematics in Engineering (Year: 2, Semester: Winter)
Master Academic Studies	Information Engineering (Year: 1, Semester: Winter)
Master Academic Studies	Information Systems Engineering (Year: 1, Semester: Winter)

General information:

Category	Theoretical-methodological
Scientific or art field	Inženjerstvo informacionih sistema Teorijska i primenjena matematika Information-Communication Systems
ECTS	4

The goal of the course is to introduce the students to the basic concepts of design and use of deep neural networks – systems whose architecture is based on the architecture of primate central nervous system. The students will understand the basic concepts of neural networks and learn how to use deep learning software packages to develop artificial intelligence systems.

Upon completing the course, the students will have the knowledge and skills that will enable them to efficiently use deep learning to solve practical problems in the domain of IT. They will also receive practical training in developing applications using the Caffe framework for neural network modeling and training.

The course will cover the following topics: basic concepts of I and II generation neural networks, data representation (coding) in neuromorphic systems, basic methods of supervised and unsupervised learning in these systems, Deep Learning and applications of neuromorphic systems to the problem of analyzing large amounts of multimedia data. The theory will be accompanied with practical training in developing and training machine learning methods in the Caffe environment.

Lectures and labs, tests and an individual assignment (project). The labs will focus on enabling the students to use Caffe to develop machine learning models for intelligent computer systems. The students’ knowledge of the theory will be evaluated using tests. The individual assignment will consist of the practical implementation of deep learning artificial intelligence system of suitable complexity.

Authors	Title	Year	Publisher	Language
Goodfellow, I., Bengio, Y., Courville, A.	Deep Learning	2017	MIT Press, Cambridge	English
Avicennasis et al.	Artificial Neural Networks	2014	WikiBooks	English
Raul Rojas	Neural Networks	1996	Springer-Verlag, Berlin	English