Neural Networks

Course Code: Ν2-6070Α
Weekly Duty 4 (2Th + 2L)
ECTS: 5
Typical Semester: 6th
Course Category: Specialty Course
Prerequisites:  

Learning Outcomes

Upon successful completion of this course the student will be able to:

  • understand the differences between the strict Von-Neumann architecture and the relatively flexible architecture of a neural network
  • understand the difference between the algorithmic way of solving the problems of classical artificial intelligence and the corresponding inductive learning procedure of artificial neural networks
  • understand the difference between supervised and unsupervised learning
  • explain the operation/functionality of various neural paradigms
  • select an appropriate neural model according to the problem to be solved
  • understand the technical possibilities, the advantages and the limitations of the various learning and/or self-organizing systems

Course Content

Introduction. Historical review. Model of artificial neuron. Neural networks training. Supevised Learning: Associative Memories (Correlation Matrix Memory – CMM, Generalized Inverse Memory – GIM), ADALINE, Hamming Network/MAXNET, Binary Hopfield Network, Perceptron, Multilayer Perceptrons and the back-propagation algorithm, Radial Basis Function (RBF) Networks, The LVQ algorithm. Unsupevised Learning: Kohonen network (Self-organizing Maps), ART (Adaptive Resonance Theory). Neurocomputers. Neural network implementations in systolic architectures. Applications.

Laboratory: Training in the environment of Matlab and the Neural Network Toolbox. Implementation of associative memories (CMM, GIM) to solve pattern recognition problems. Implementation of the ADALINE network for signal tracking. Implementation of Hopfield net on a synthetic problem. Implementation of a simple perceptron for solving a linearly separable problem. Implementation of multilayer neural network trained with the back-propagation and Levenberg-Marquardt algorithms for solving non-linear problems. Implementation of LVQ and Self-organizing Maps on synthetic problems.

Literature
  1. HaykinS., Νευρωνικά Δίκτυα και Μηχανική Μάθηση, Εκδόσεις Παπασωτηρίου, 2010.
  2. Διαμαντάρας Κ., Τεχνητά Νευρωνικά Δίκτυα, Εκδόσεις Κλειδάριθμος, 2007.
  3. Ρίζος Γ., Τεχνητά Νευρωνικά Δίκτυα: Θεωρία και Εφαρμογές, Εκδόσεις Νέων Τεχνολογιών, 1996.
  4. Bishop M., Neural Networks for Pattern Recognition, Clarendon Press, 1997.
  5. Lin C-T., Lee C.S.G., Neural Fuzzy Systems: A Neuro-Fuzzy Synergism to Intelligent Systems, Prentice Hall, 1996.
  6. Kohonen T., Self-Organizing Maps, Springer Verlag, 1995.
  7. Haykin S., Neural Networks: A Comprehensive Foundation, McMillan, 1994.
  8. Hertz J., Krogh A., Palmer R., Introduction to the Theory of Neural Computation, Addison Wesley, 1991.

Internationalisation I18n