MIWAI'2014
The 8th Multi-Disciplinary International Workshop on Artificial Intelligence
1
2
3
4
5
6
7- Home
- Call for Papers
- Committees
- Submission
- Technical Programme
- Speakers
- Tutorials
- Accepted Papers
- Registration
- Venue & Accommodation
- Sponsors
- Contacts
- Location
Publication
Proceedings of MIWAI 2014 will be published in Springer LNAI series.
Flags Counter started on April 25, 2014
Speakers
Prof. C. A. Murthy (Keynote Speaker)
 |
Prof. C. A. Murthy Machine Intelligence Unitarianize Indian Statistical Institute 203 Barrackpore Trunk Road Kolkata - 700 108 India Email: murthy@isical.ac.in http://www.isical.ac.in/~murthy/ More... |
Prof. P. S. Sastry (Invited Speaker)
 |
Prof. P. S. Sastry Room No 235 Department of Electrical Engineering Indian Institute of Science Bangalore-560012 India Email: sastry@ee.iisc.ernet.in http://minchu.ee.iisc.ernet.in/new/people/faculty/pss/ More... |
Brief Biography: Professor P.S Sastry received B.Sc.(Hons.) in Physics from IIT, Kharagpur, in 1978 and B.E. in Electrical Communications Engineering and Ph.D. from Department of Electrical Engineering, from IISc, Bangalore in 1981 and 1985 respectively. Since 1986, he has been a faculty member in the department of Electrical Engineering, Indian Institute of Science, Bangalore, where currently he is a professor. He has held visiting positions at University of Massachusetts, Amherst, University of Michigan, Ann Arbor and General Motors Research Labs, Warren, USA. He is an Associate Editor of IEEE Transactions on Systems, Man and Cybernetics since 2001 and IEEE Transactions on Automation Science and Engineering (2007-10). He is a recipient of Sir C.V.Raman Young Scientist award, from Govt of Karnataka, Hari Om Ashram Dr. Vikram Sarabhai research award from PRL and Most Valued Colleague award from General Motors Corporation. He is a Fellow of Indian National Academy of Engineering and a Senior Member of IEEE. He has around 45 high quality publications. He also has 3 patents to his credit. His research interests include Machine Learning, Pattern Recognition, Data Mining and Computational Neuroscience.
Title: Learning classifiers under label noise
Abstract: Supervised learning of pattern classifiers is a basic problem in Machine Learning. One uses a training set of patterns vectors with known class labels for learning the classifier. In many applications the labels provided for the training patterns may be incorrect and this is termed as label noise. This problem is important in many applications today where the labels may have been obtained through, e.g., crowd sourcing. Obviously it is desirable to have learning algorithms that can learn good classifiers even though the training set may be corrupted with label noise. Many approaches have been proposed for identifying and/or guarding against training examples with noisy labels. In the first part of this talk we will review some of the approaches based on heuristics and also some of the recent approaches that are based on some interesting statistical principles. In the second part of the talk we concentrate of one generic approach to classifier learning, namely, risk minimization. We present some of our recent results on making risk minimization robust to label noise. We consider the cases of both uniform and non-uniform label noise. In the uniform noise case, the probability of class label being incorrect is same for all feature vectors while under the non-uniform label noise, the probability of the class label being incorrect can be a function of the feature vector. We present some sufficient conditions on the loss function that would make risk minimization robust to label noise. While none of the popular convex loss functions such as the hinge loss (which is used in SVM), the exponential loss (which is used in AdaBoost) etc. satisfy the sufficient conditions, we show that ramp loss, sigmoid loss and the 0-1 loss satisfy the sufficient condition. We also discuss some algorithmic issues in minimizing empirical risk under the noise-robust loss functions. We present some empirical results that demonstrate the relevance of these theoretical results.
Dr. B. S. Daya Sagar (Invited Speaker)
 |
Dr. B. S. Daya Sagar Associate Professor & Head Systems Science and Informatics Unit (SSIU) Indian Statistical Institute--Bangalore Centre 8th Mile, Mysore Road, R. V. College P.O Bangalore-560059 India Email: bsdsagar@isibang.ac.in http://www.isibang.ac.in/~bsdsagar/ More... |
Dr. Biplav Srivastava (Invited Speaker)
 |
Dr. Biplav Srivastava Senior Researcher, IBM Master Inventor IBM Research, ISID Campus, Plot No. 4, Block C, Institutional Area, Vasant Kunj New Delhi - 110 070 India Email: sbiplav@in.ibm.com http://www.research.ibm.com/people/b/biplav/ More... |
Dr. P. Krishna Reddy (Tutorial Speaker)
 |
Dr P.Krishna Reddy Program Director, ITRA Agriculture & Food, DietY, Govt. of India and Professor International Institute of Information Technology Hyderabad (IIIT-H) Gachibowli, Hyderabad, Telangana State 500032 India E-mail: pkreddy@iiit.ac.in http://www.iiit.ac.in/~pkreddy/ More... |
Dr. Chattrakul Sombattheera (Tutorial Speaker)
 |
Dr. Chattrakul Sombattheera Faculty of Informatics Mahasarakham University Khamreang Sub-District Kantarawichai District Maha Sarakham 44150 Thailand Email: chattrakul.s@msu.ac.th More... |