Neural Networks

Overview | Contents | Prerequisites | Course materials (in romanian) | Homework (in romanian) |
Projects( in romanian) | Bibliography | Links

Overview:

This is a one-semester course for the third year students of the Computer science section. The aim of the course is to introduce the main principles of neural computation.

Basic notions: computational neurons, artificial neural networks, architectures, functioning, learning.
Single-layer feedforward networks with supervised learning (the perceptron's, Widrow-Hoff's and delta algorithms). Applications in classification.
Multi-layer feedforward networks with supervised learning (the BackPropagation algorithm). Applications for pattern recognition and function approximation.
Single-layer feedforward networks with unsupervised learning (correlative and competitive algorithms). Applications for principal component analysis and clustering. Self-organizing maps.
Radial basis function networks. Applications in function approximation.
Recurrent neural networks. Applications for associative and optimization tasks, for time series prediction and image processing.
Genetic algorithms and genetic programming. Applications for optimization problems.

Prerequisites:

Programming Languages
Basic Linear Algebra
Numerical Analysis
Statistics

Course materials (in romanian)

Curs1(PDF): Introducere in calculul inteligent (folii)
Curs2(PDF) : Structura si proiectarea retelelor neuronale (folii)
Curs3-4 (PDF) Retele feed-forward cu un nivel (folii (C3), folii (C4))
Curs5 Retele feed-forward cu nivele ascunse. Algoritmul BackPropagation (folii)
Curs6 (PDF): Variante ale algoritmului BackPropagation (folii)
Curs 7 (PDF): Retele cu functii radiale de transfer (folii)
Curs8-9(PDF): Retele neuronale cu invatare nesupervizata. Retele Kohonen. Retele neuronale pentru analiza in componente principale ( Curs10 (PDF)) (folii curs 8, folii curs 9)
Curs11-12 (PDF): Retele recurente (folii curs 10, folii curs 11)
Curs 13: Algoritmi evolutivi si utilizarea in proiectarea retelelor neuronale

Lab1: Introducere in Matlab Neural NetworksToolbox.
Lab2: Simularea perceptronului. Aplicatii in clasificarea liniar separabila (clasificarea unor simboluri: citireSablon.m, clasificare.m).
Lab3: Simularea retelelor feedforward cu mai multe nivele (regresie liniara si neliniara: aproximare.m)
Lab4: Simularea retelelor RBF. Aproximare ( aproximareRBF.m, aproximareRBFin.m) si predictie ( predictieBP.m, predictieRBF.m, date.dat)
Lab5: Simularea retelelor neuronale cu invatare bazata pe procese de competitie. Aplicatie: comparare intre algoritmul kmeans si o retea simpla cu invatare competitiva ( kmeans.m, clusteringWTA.m)
Lab 6: Simularea retelelor Kohonen. Aplicatie: Kohonen_TSP.m
Lab 7: Simularea retelelor recurente (Hopfield, Elman). Aplicatii: maHopfield.m, elman1.m,

Teme:

Tema1: Recunoastere de caractere cu retele feedforward cu un nivel
Tema2: Implementarea unei retele feedforward cu un nivel ascuns. Rezolvarea unei probleme de predictie. (Date de test)
Tema3: Implementarea unei re'tele adaptive pentru rezolvarea problemelor de clustering.

Proiecte :

Proiecte optionale: proiecte_2006.pdf

Bibliography:

Kosko, B. Neural Networks and Fuzzy Systems: A Dynamical Systems Approach to Machine Intelligence. Prentice Hall, Englewood Cliffs, 1992.
Kung, S.Y. , Digital neural networks, Prentice Hall, 1993.
Dumitrescu, D., Costin, H. Retele neuronale; teorie si aplicatii. Teora, Bucuresti, 1996.
Dumitrecu D., Algoritmi genetici si strategii evolutive, Microinformatica Cluj, 2000.
Müller, B., Reinhardt, J. Neural Networks – An Introduction. Springer-Verlag, Berlin, 1990.
Kohonen, T. Self–Organization and Associative Memory. Series in Information Sciences, vol. 61, Springer–Verlag, Berlin, 1989.
Wasserman, P. Neural Computing – Advanced Methods. Van Nostrand Reinhold Inc., Computer Science Series, 1993.
Masters, T. Practical Neural Networks Recipes in C++. Academic Press, Boston, 1993.
Ripley, B.D., Pattern Recognition and Neural Networks, Cambridge University Press, 1996.

Carti existente la BCUT (http://www.bcut.ro/):

GELENBE, Erol (ed.), NEURAL networks, 1992.
MASTERS, Timothy, Advanced algorithms for neural networks, 1995.
HAGAN, Martin T., Neural network design, 1995.
BELTRATTI, Andrea, Neural networks for economic and financial modeling, 1996.
SKAPURA, David M., Building neural networks, 1996.
ELLACOTT, Stephen W., MATHEMATICS of neural networks, 1997.
RUAN, Da (ed.), INTELLIGENT hybrid systems, 1997
MITCHELL, Melanie, An introduction to genetic algorithms, 1996.
KRUSE, Rudolf, Fuzzy-systems in computer science, 1994.
TERANO, Toshiro (ed.), FUZZY systems theory and its applications, 1992.
ISHIKAWA, Akira, Analysis and evaluation of fuzzy systems, 1995.
MORABITO, F.C. (ed.), ADVANCES in intelligent systems, 1997.
DUBOIS, Didier, Fuzzy sets and systems, 1980.
LOOTSMA, Freerk A., Fuzzy logic for planning decision ma..., 1997.

Links:

Go back to index