Hybrid intelligent systems: design, challenges and applications

Prof. Ajith Abraham
VŠB-Technical University of Ostrava, Ostrava-Poruba, Czech Republic

20 hours, 5 credits

May 9 - May 12, 2011

Dipartimento di Ingegneria dell'Informazione: Elettronica, Informatica, Telecomunicazioni, Largo Lucio Lazzarino, meeting room

Contacts: Prof. Francesco Marcelloni

Objectives

The objective of this series of lessons is to provide a solid foundation of some of the modern computational intelligent tools, namely neural networks, fuzzy systems, evolutionary algorithms and swarm intelligence, and then focus on constructing adaptive / complex hybrid approaches by combining some of these fundamental tools.

At the end, students should know the basic components of modern intelligent systems, know some interesting application fields for computational intelligence, should be able to design their own intelligent system and do the implementation of (parts of) such a system. The lessons aim to give students more breadth (theoretical knowledge), more depth (hands-on experience), learn a further framework for discriminating and interrelating approaches to computational intelligence, survey the increasingly broad range of applications, take on (larger) programming projects.

Contents

• What is computational intelligence?
• Fuzzy sets, fuzzy reasoning using fuzzy if-then rules
• Artificial Neural Networks: advanced learning schemes and applications
• Evolutionary Computation and Swarm Intelligence
  
• Hybrid Intelligent Systems
• Why hybrid systems?
• Hybrid system architectures
• Cooperative hybrid systems
• Concurrent hybrid systems
• Modular hybrid systems
• Integrated hybrid systems
• Architecture of a neuro-fuzzy system
• Evolutionary neural networks
• Evolutionary fuzzy systems
• Application case study

Laboratories

Hybrid Intelligent systems are now in use for solving many real world problems in engineering, science, commerce etc. The laboratory sessions demonstrates practical implementations of modern computational intelligence tools namely neural networks using advanced learning methods, fuzzy expert systems, evolutionary algorithms and hybrid approaches. Students will have hands on opportunities to develop their own ways of implementing some hybrid approaches to solve some problems or use it for real world applications.