1. Comprehending the principles underlying the various processes involved in fuzzy systems: linguistic descriptions, fuzzy IF/THEN rules, rules of inference.
2. For a given problem the student should be able to formulate a suitable fuzzy rule base, select the implication operators, the fuzzification and the de-fuzzification strategies.
3. In regard to control tasks, the student should be able to develop the proper controller structure, formulate its fuzzy rule base and select suitable gain parameters for optimal system’s response.
4. In regard to fuzzy modeling for prediction tasks, design the appropriate fuzzy model and implement the parameters learning algorithm.
5. Comprehending the principles of fuzzy clustering and deal with applications to data classification.
6. Understanding the principle, structures and learning techniques involved in NNs.
7. Examining NNs of various structures, the relevant training techniques, and their applications to control, modeling and classification problems.
8. Integration of fuzzy systems and neural network models.
Course Content (Syllabus)
Fuzzy Systems: Fuzzy sets, properties, fuzzy operators and membership functions. Resolution and extension theorems, a-cuts, fuzzy union, intersection and complement.
Fuzzy relations, operations between fuzzy relations, fuzzy relation composition, fuzzy set-relation composition.
Fuzzy If/Then rules and implication functions. Fuzzy rule bases, compositional rule of inference, fuzzification and defuzzications structures.
Fuzzy controllers, controller structures. Design of fuzzy controllers FZ-PI, FZ-PD and FZ-PID. Comparative gain tuning, experimental results of fuzzy controllers.
Fuzzy TSK models and models with crisp outputs. Training algorithms for adaptive neuro-fuzzy networks. Fuzzy clustering algorithms, the Fuzzy C-means method.
Neural Networks (NNs): Perceptron model and learning rules. Supervised, unsupervised and reinforcement learning techniques. Single layer and multi-layer networks. The back-propagation algorithm. RBF networks and equivalence to fuzzy systems. Self-organizing networks SOFM. The learning vector quantization (LVQ) algorithm. Applications of NNs to prediction and control tasks.
Integration between fuzzy and NN systems, neuron-fuzzy models.
Course Bibliography (Eudoxus)
1. Υπολογιστική Νοημοσύνη και Εφαρμογές, Ι. Μπούταλης, Γ. Συρακούλης, Εκδόσεις: Γ. ΣΥΡΑΚΟΥΛΗΣ, 2010, ISBN: 978-960-93-2008-5
2. Eισαγωγή στην Ασαφή Λογική (Fuzzy Logic)», Γ. Θεοδώρου, Εκδόσεις: ΤΖΙΟΛΑ, 2010, ISBN: 978-960-418-218-3.
3. Neural Fuzzy Systems, C.-T. Lin and C. S. G. Lee, Prentice Hall, Inc., 1996.
4. Neuro-fuzzy and Soft Computing. A computational approach to learning and machine intelligence, J.-S. R. Yang, C.-T. Sun, and E. Mizutani, Prentice Hall, NJ, 1997.