Finite Precision Analysis of Support Vector Machine Classification in Logarithmic Number Sy.pdf

Finite Precision Analysis of Support Vector Machine Classification in Logarithmic Number Systems Faisal M. Khan, Mark G. Arnold and William M. Pottenger Computer Science and Engineering, Lehigh University, Bethlehem, Pennsylvania {fmk2,maab,billp}@ Abstract Support Vector Machines (SVMs), a representative kernel-based machine-learning technique esp ecially suited to high-dimensional data [13], [19], [20], [24]. In this paper we present an analysis of the mini- As noted, significant progress has been made in mal hardware precision required to implement Sup- the software domain for modeling and replicating the port Vector Machine (SVM) classifi cation within a natural processes of learning, adapting and decision Logarithmic Number System arch itecture. Support making for intelligent data analysis. Unfortunately, Vector Machines are fast emerging as a powerful ma- such solutions require significant resources for execu- chine-learning tool for pattern recognition, decision - tion and m