Courses – Fakultet tehničkih nauka u Novom Sadu

Subject: Self-Learning and Adaptive Algorithms (17.AUN54)

General information:

Category	Scientific-professional
Scientific or art field	Automatic Control and System Engineering
ECTS	4

Prepares students for solving basic problems in analysis, synthesis and implementation of self-learning and adaptive systems in decision support problems and elsewhere. Introduces them to appropriate literature and prepares them for individual work in the field.

The students will acquire basic knowledge in the field of self-learning and adaptive systems and algorithms. They will be trained to select the appropriate algorithms, select meta-parameters, and implement it on appropriate platform.

1. Basic notions on decision support systems, machine learning, adaptive and self-learning systems. 2. Finite Markov decision processes. 3. Basic methods of exactly solving finite decision problems. 4. Limitations of exact methods and necessity for introducing approximations - examples and case studies. 5. Linear regression and classification - Least squares. 6. Adaptive parameter estimation - Recursive least squares and Kalman filter. 7. Non-linear regression and classification. Adaptive estimation of parameters in non-linear models. 8. Artificial neural networks (ANN) as an example of non-linear regression and classification. Backpropagation algorithm. 9. Stochastic gradients and steepest descent for ANN training. 10. Adaptive estimation of parameters in linear models. 8. Linear predictors and adaptive linear predictors.

Lectures. Computer-based exercises. Consults. Projects.

Authors	Title	Year	Publisher	Language
I. Moreels and J. Willem	Adaptive Systems - An Introduction	1996	Birkhauser	English
Ioannou, P.A.	Adaptive Systems with Reduced Models	1983	Springer-Verlag, Berlin	English
V. Vapnik	Statistical Learning Theory	1998	John Willey and Sons	English
C. Gres	Complex and Adaptive Systems	2008	Springer	English
A. Zaknich	Principles of Adaptive Filters and Self Learning Systems	2005	Springer	English
Ruchard S. Sutton, Andrew G. Barto	Reinforced Learning - An Introduction	2017	MIT Press	English