Improvements to Platt's SMO Algorithm for SVM Classifier Design
S. Sathiya KeerthiDepartment of Mechanical and Production Engineering, National University of Singapore, Singapore-119260Shirish ShevadeDepartment of Computer Science and Automation, Indian Institute of Science, Bangalore-560012, IndiaChiranjib BhattacharyyaDepartment of Computer Science and Automation, Indian Institute of Science, Bangalore-560012, IndiaK. R. K. MurthyDepartment of Computer Science and Automation, Indian Institute of Science, Bangalore-560012, India
2001en
ABI
Abstract
This article points out an important source of inefficiency in Platt's sequential minimal optimization (SMO) algorithm 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 modifications of SMO. These modified algorithms perform significantly faster than the original SMO on all benchmark data sets tried.
Identifiers
Citations and references
Cited by 20 references