A Comparison of Quadratic Programming Solvers in Support Vector Machines Training

Autor:	S. Deris, N. M. Zaki, K. K. Chin
Rok vydání:	2012
Předmět:	Support vector machine Theoretical computer science Computer science General Engineering Quadratic programming Arithmetic
Zdroj:	Jurnal Teknologi.
ISSN:	2180-3722 0127-9696
DOI:	10.11113/jt.v39.461
Popis:	Penyelesaian atur cara kuadratik yang sangat besar diperlukan untuk melatih Support Vector Machine. Tiga cara penyelesaian atur cara kuadratik yang berbeza telah digunakan untuk melaksanakan latihan Support Vector Machine bagi mengkaji keberkesanannya ke atas Support Vector Machine. Prestasi bagi kesemua penyelesaian telah dikaji dan dianalisis dari segi masa pelaksanaan dan kualiti penyelesaian. Kaedah praktikal untuk mengurangkan masa latihan tersebut telah dikaji sepenuhnya. Kata kunci: Support vector machines, atur cara kuadratik Training a Support Vector Machine requires the solution of a very large quadratic programming problem. In order to study the influence of a particular quadratic programming solver on the Support Vector Machine, three different quadratic programming solvers are used to perform the Support Vector Machine training. The performance of these solvers in term of execution time and quality of the solutions are analyzed and compared. A practical method to reduce the training time is investigated. Key words: Support vector machines, quadratic programming
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::899d545d89e79051697727769d0be84f https://doi.org/10.11113/jt.v39.461 Zobrazit plný text záznamu