1-AIN-480 Neural Networks

Evaluation during the course: individual projects

Concluding evaluation: exam

Subject aim: Provide introduction to principles of neural computation and the usability of various neural network models in solving various tasks (pattern recognition, classification, time series prediction, autoassociative memory and others).

Brief curriculum of the subject:

Introduction to artificial neural networks (connectionism), brief history, characteristic features.

Learning in neural networks, learning algorithms, generalization to new data.

Pattern recognition and classification usind feedforward models.

Linear models, autoassociative memory, memory capacity.

Self-organized learning (unsupervised), feature extraction.

Topographic mapping, visualization of high-dimensional data.

Hybrid models (supervised and unspervised learning).

Recurrent models: learning temporal structure in data.

Hopfield model: memorized patterns as attractors in state space.

Connectionizm in cognitive science.

Literature:

Kvasnička V., Beňušková Ľ., Pospíchal J., Farkaš I., Tiňo P. a Kráľ A. kol.: Introduction into theory of neural networks (in Slovak). IRIS Bratislava, 1997.

Haykin S.: Neural networks. MacMillan Press, 2000 (2nd edition).

Language in which the subject is taught: Slovak

Date of the last sheet revision: 24.8.2007