Invited Speakers

Nikhil R. Pal
Indian Statistical Institute
Calcutta, India

Making neural networks smart enough to say don’t know — the open world classification problem

Abstract. In an open-world situation, a test data point may come from an unknown class and a trained network will fail. In a closed-world situation, a test data point may come from outside the “sampling window” and here the network should not make a decision, but it will. Sometimes, it may even assign an unrealistic class label. The statistical properties of one or more class may also change significantly with time and in this case also the network may fail.

These problems are different in nature, but all have one thing common: these problems arise when a network makes a decision where it should not. In the literature there are attempts to address each of these problems separately. Here we shall develop some methods that can address all the problems in a single framework. In this context, we shall briefly discuss some probabilistic methods and how the Extreme Value Theorem is used to address the open world classification problem. Then we shall present methods to model the complement world of the sampling window and use that to solve these problems – this approach can be viewed as making the network “smart” enough to say don’t know in the right context. Some theoretical results will also be highlighted. Does it mean that we have solved the problem? No! We consider it still an “open problem”. We shall mention some of the areas where further research focus is needed.

Nikhil R. Pal is a Professor in the Electronics and Communication Sciences Unit of the Indian Statistical Institute. His current research interest includes brain science, computational intelligence, machine learning and data mining.

He was the Editor-in-Chief of the IEEE Transactions on Fuzzy Systems for the period 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, International Journal of Neural Systems, Fuzzy Sets and Systems, IEEE Transactions on Fuzzy Systems and the IEEE Transactions on Cybernetics.

He is a recipient of the 2015 IEEE Computational Intelligence Society (CIS) Fuzzy Systems Pioneer Award. He has given many plenary/keynote speeches in different premier international conferences in the area of computational intelligence. He is a Distinguished Lecturer of the IEEE CIS (2010–2012, 2016–2018) and was a member of the Administrative Committee of the IEEE CIS (2010–2012). He has served as the Vice-President for Publications of the IEEE CIS (2013–2016). He is serving as the President of the IEEE CIS (2018–2019).

He is a Fellow of the National Academy of Sciences of India, International Fuzzy Systems Association, The World Academy of Sciences, and a Fellow of the IEEE, USA.