2-5 February 2016

Canberra, Australia


Nikhil R. Pal

Nikhil R. Pal Biography: Nikhil R. Pal ( is an INAE Chair Professor in the Electronics and Communication Sciences Unit of the Indian Statistical Institute. His current research interest includes bioinformatics, brain science, fuzzy logic, neural networks, machine learning, and data mining. He was the Editor-in-Chief of the IEEE Transactions on Fuzzy Systems (January 2005-December 2010). He has served/been serving on the editorial /advisory board/ steering committee of several journals including the International Journal of Approximate Reasoning, Applied Soft Computing, Fuzzy Sets and Systems, Fuzzy Information and Engineering : An International Journal, IEEE Transactions on Fuzzy Systems and the IEEE Transactions on Systems Man and Cybernetics B (currently IEEE Transactions on Cybernetics).

He is a recipient of the 2015 Fuzzy Systems Pioneer Award. He has given many plenary/keynote speeches in different premier international conferences in the area of computational intelligence. He has served as the General Chair, Program Chair, and co-Program chair of several conferences. He was a Distinguished Lecturer of the IEEE Computational Intelligence Society (CIS) and was a member of the Administrative Committee of the IEEE CIS. At present he is the Vice President for Publications of the IEEE CIS.

He is a Fellow of the National Academy of Sciences, India; the Indian National Academy of Engineering; the Indian National Science Academy, the International Fuzzy Systems Association (IFSA), and IEEE, USA.

Title: Dimensionality Reduction: Neural and Neuro-Fuzzy Frameworks

Abstract:Successful design of any “intelligent system” demands use of an appropriate set of features. Ideally a useful set of features should select necessary features, discard derogatory features and indifferent features, and it should have a controlled level of redundancy. Complete elimination of redundancy is not desirable as then the system may not be able to tolerate any measurement error. The feature selection problem can be generalized to sensor selection where a sensor is responsible for a set of features. In this talk, first we shall discuss how neural networks can be effectively used for structure-preserving dimensionality reduction. This approach does not use class labels and since it preserves the “inherent structure” in the high dimensional data, it is quite effective for most applications, in particular, for data visualization. This will be followed by how neural networks and neuro-fuzzy systems can be adapted to select useful features discarding derogatory and indifferent features. The neural model will then be enhanced to deal with the problem of sensor selection. Finally, we shall present how the neural networks can select features (as well as sensors) with a control on the level of redundancy in the set of selected features (sensors).