Memory-efficient large-scale linear support vector machine


Alrajeh, Abdullah, Takeda, Akiko and Niranjan, Mahesan (2014) Memory-efficient large-scale linear support vector machine At The 7th International Conference on Machine Vision, Italy. 19 - 21 Nov 2014.

Download

[img] PDF icmv14.pdf - Author's Original
Download (300kB)
[img] PDF svm.pdf - Other
Download (309kB)

Description/Abstract

Stochastic gradient descent has been advanced as a computationally efficient method for large-scale problems. In classification problems, many proposed linear support vector machines as very effective classifiers. However, they assume that the data is already in memory which might not be always the case. Recent work suggests a classical method that divides such a problem into smaller blocks and then solves the sub-problems iteratively. We show that a simple modification of shrinking the dataset early will produce significant saving in computation and memory. We further ?nd that on problems larger than previously considered, our approach is able to reach solutions on top-end desktop machines while competing methods cannot.

Item Type: Conference or Workshop Item (Paper)
Venue - Dates: The 7th International Conference on Machine Vision, Italy, 2014-11-19 - 2014-11-21
Related URLs:
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Organisations: Electronics & Computer Science
ePrint ID: 368208
Date :
Date Event
November 2014Published
Date Deposited: 20 Aug 2014 15:26
Last Modified: 17 Apr 2017 13:11
Further Information:Google Scholar
URI: http://eprints.soton.ac.uk/id/eprint/368208

Actions (login required)

View Item View Item