alaidiText Book
Jang, J.-S. R., Sun, C.-T. and Mizutani, E. Neuro-Fuzzy and Soft Computing: A Computational Approach to Learning and Machine Intelligence. Prentice Hall, 1997.
Sivanandam, S. N. and Deepa, S. N. Principles of Soft Computing. 3rd Edition, Wiley, 2018.
Ross, T. J. Fuzzy Logic with Engineering Applications. 4th Edition, Wiley, 2016.
Mitchell, M. An Introduction to Genetic Algorithms. MIT Press, 1998.
Syllabus
| Week | Date | Covered Topics | Comments |
|---|---|---|---|
| 1 | Course organization | Aims, assessment, project brief | |
| 2 | What is Soft Computing? Hard vs. soft computing; tolerance for imprecision and uncertainty; the three pillars — fuzzy logic, neural networks, evolutionary computation; where each is the right tool | Jang ch. 1 | |
| 3 | Fuzzy sets. Classical vs. fuzzy sets; membership functions and their shapes; support, core, α-cuts; convexity and normality | Jang ch. 2 | |
| 4 | Operations and relations. Union, intersection, complement; t-norms and t-conorms; fuzzy relations; max–min composition; the extension principle | Jang ch. 2–3 | |
| 5 | Fuzzy logic. Linguistic variables and hedges; fuzzy propositions; fuzzy if–then rules; approximate reasoning | Jang ch. 4 | |
| 6 | Mamdani inference. Fuzzification, rule evaluation, aggregation; defuzzification methods (centroid, bisector, MOM) compared | Jang ch. 4 | |
| 7 | Sugeno (TSK) inference and Tsukamoto; when each is preferable; designing a fuzzy controller end to end | Lab: fuzzy controller | |
| 8 | Review and Exam 1 — fuzzy systems | Exam 1 | |
| 9 | Neural networks: foundations. Biological motivation; McCulloch–Pitts neuron; perceptron and its convergence theorem; the linear separability limit | Jang ch. 9 | |
| 10 | Multilayer perceptron. Backpropagation derived; activation functions; learning rate and momentum; overfitting, validation, early stopping | Jang ch. 9 | |
| 11 | Other architectures. Radial basis function networks; self-organizing maps (Kohonen); Hopfield networks and associative memory | Jang ch. 9–10 | |
| 12 | Neuro-fuzzy systems. ANFIS: architecture, hybrid learning (least squares + gradient descent); what the fusion buys you | Jang ch. 12 | |
| 13 | Genetic algorithms. Encoding; fitness; selection (roulette, tournament, rank); crossover and mutation; elitism; the schema theorem | Project — phase 1 due | |
| 14 | Beyond the simple GA. Convergence and premature convergence; parameter tuning; multi-objective optimisation and NSGA-II | Mitchell ch. 1–2 | |
| 15 | Swarm intelligence. Particle swarm optimisation; ant colony optimisation; comparison with GA on the same problem | Lab: PSO vs. GA | |
| 16 | Hybrid systems and applications; project presentations | Project presentation |