Tutorial 1: Computational Intelligence Techniques with Applications

Abstract: Computational intelligence techniques find applications in many diverse fields including medicine, control, imaging and so on. Bioinformatics and medical image analysis are important areas in medicine having many challenging problems related to mankind. For example, automatic screening of breast cancer is still a challenging problem to researchers and needless to say that a low cost solution can save many lives. Similarly, predicting the structure of new protein or finding of biomarkers for a specific type of cancer can save lives. There is a tremendous interest among the researchers to exploit the power of Computational intelligence tools to solve such problems. Intelligent control is another area where computational intelligence plays a very important role. This tutorial intends to provide adequate knowledge of computational intelligence tools and how these tools can be used to solve some of these challenging problems. As applications, we shall consider the following problems: identification of biomarkers (genes) for specific cancers, knowledge discovery in biological data, detection of microcalcification in mammograms, automatic designing of self-tuning controllers, etc.

Computational intelligence (CI) primarily encompasses biologically and linguistically motivated computing paradigms. Three main ingredients of CI are fuzzy computing, neuro-computing and evolutionary computing. The tutorial will be divided into four parts.

Part I: We shall introduce CI, its basic components, relevance and advantages. We shall discuss some important supervised and unsupervised models such as multilayer perceptrons, support vector machines, self-organizing maps, neural gas networks and so on. We will discuss some applications of these techniques in feature/gene selection, detection of microcalcifications from mammograms, image compression, segmentation of remote sensing data, sensor selection, intelligent control, etc.

Part II: In our day to day life we process primarily imprecise information and fuzzy sets is the most appropriate tool to model such imprecision and approximate reasoning with that. One can think neural network as hardware for the brain (low level information processing tool) and fuzzy reasoning as a higher level information processing. We shall discuss the motivation and need for fuzzy sets, and then introduce it formally along with various operations on the same. The concept of linguistic variables and linguistic values and how to deal with them will also be discussed. We shall discuss the general architecture of a fuzzy rule based systems along with different inference and defuzzification schemes. As applications we shall consider satellite image analysis, detection of bounded weak echo regions relating to severe weather phenomena, and fuzzy control.

Part III: To design either fuzzy systems or neuro-systems we often need to optimize, and sometimes such objective functions are not well behaved. Evolutionary computing, another biologically inspired computing paradigm, is an attractive tool for solving such optimization problems. In this context, we shall introduce genetic algorithm, genetic programming, particle swarm optimization and differential evolution. We will also discuss their applications in gene/feature selection and gene regulatory networks.

Part IV: Each of the three major components of CI has its own advantages and limitations. Hence, a judicious integration of two or more of these tools into one system can lead to more “intelligent systems” with better properties. In this regard, we shall discuss different hybrid systems such as neuro-fuzzy, fuzzy-genetic systems. Selected applications of these hybrid systems will also be discussed.

Biography: Professor Nikhil R. Pal received B. Sc. degree with honors in physics and Master degree in business management from the University of Calcutta. He obtained MTech and PhD degrees in computer science from the Indian Statistical Institute, Calcutta. Currently, he is a Professor in the Electronics and Communication Sciences Unit of the Indian Statistical Institute. He is a Fellow of the IEEE, USA and Indian National Academy of Engineering. He has delivered several plenary, tutorial and invited talks.

He has coauthored a book titled "Fuzzy Models and Algorithms for Pattern Recognition and Image Processing", Kluwer Academic Publishers, 1999; co-edited four volumes titled "Advances in Pattern Recognition and Digital Techniques", ICAPRDT'99, Narosa; "Advances in Soft Computing", AFSS 2002, Springer Verlag; "Neural Information Processing", ICONIP 2004, Springer Verlag; "Advanced echniques in data mining and knowledge discovery", Springer Verlag, 2005; and edited a book titled "Pattern Recognition in Soft Computing Paradigm", World Scientific, 2001.

He is the Editor-in-Chief of IEEE Transactions on Fuzzy Systems. He is an Associate Editor of the IEEE Transactions on Systems, Man and Cybernetics-B. He serves on the editorial/advisory board of the International Journal of Approximate Reasoning, International Journal of Hybrid Intelligent Systems, Neural Information Processing - Letters and Reviews, International Journal of Knowledge-Based Intelligent Engineering Systems, Iranian Journal of Fuzzy Systems, Fuzzy Sets and Systems, and International Journal of Neural Systems.

He is a Steering Committee member of the journal "Applied Soft Computing", Elsevier Science. He was the president of and currently is serving as a governing board member of the Asia Pacific Neural Net Assembly. He was the Program Chair of the 4th International Conference on Advances in Pattern recognition and Digital Techniques, Dec. 1999, Calcutta, India. He was the General Chair of 2002 AFSS International Conference on Fuzzy Systems, Calcutta, 2002 and the 11th International Conference on Neural Information Processing, ICONIP 2004. He was a co-program chair of 2005 IEEE International Conference on Fuzzy Systems and is a co-program chair of 2006 IEEE International Conference on Fuzzy Systems.