S. S. Keerthi S. K. Shevade C. Bhattacharyya

Learn More
This paper points out an important source of ineeciency in Smola and Schh olkopf's Sequential Minimal Optimization (SMO) algorithm for SVM regression that is caused by the use of a single threshold value. Using clues from the KKT conditions for the dual problem, two threshold parameters are employed to derive modiications of SMO for regression. These(More)
The project of Analysis on the Wire (AoW) at Brookhaven National Laboratory has the goal of executing generic computations in the network fabric, "on the wire," to sup-port early decision making. We are research-ing, developing, and evaluating hardware and software mechanisms and middleware to be used for data analysis on the wire. We further pursue to(More)
1 A revised version of this report is under preparation for submission to a journal. We welcome any comments and suggestions for improving this report. Abstract In this paper we give a new, fast iterative algorithm for support vector machine (SVM) classiier design. The basic problem treated is one that does not allow classiication violations. The problem is(More)
This paper describes current efforts to architect, research, develop, and test a next-generation, high-bandwidth network monitoring framework designed to handle the rigors of large scientific feeds. This framework will be capable of transparently capturing and analyzing network traffic in real time so as to enable early and rapid response to potential(More)
  • 1