A4 Refereed article in a conference publication
Speedy Local Search for Semi-Supervised Regularized Least-Squares
Authors: Gieseke F, Kramer O, Airola A, Pahikkala T
Editors: Joscha Bach, Stefan Edelkamp
Conference name: 34th Annual German Conference on AI
Publication year: 2011
Journal: Lecture Notes in Computer Science
Book title : KI 2011: Advances in Artificial Intelligence: 34th Annual German Conference on AI, Berlin, Germany, October 4-7,2011. Proceedings
Journal name in source: KI 2011: ADVANCES IN ARTIFICIAL INTELLIGENCE
Journal acronym: LECT NOTES ARTIF INT
Series title: Lecture Notes in Computer Science
Volume: 7006
First page : 87
Last page: 98
Number of pages: 2
ISBN: 978-3-642-24454-4
eISBN: 978-3-642-24455-1
ISSN: 0302-9743
DOI: https://doi.org/10.1007/978-3-642-24455-1_8
Self-archived copy’s web address: http://research.utu.fi/converis/portal/Publication/3023302
In real-world machine learning scenarios, labeled data is often rare while unlabeled data can be obtained easily. Semi-supervised approaches aim at improving the prediction performance by taking both the labeled as well as the unlabeled part of the data into account. In particular, semi-supervised support vector machines favor decision hyperplanes which lie in a "low-density area" induced by the unlabeled patterns (while still considering the labeled part of the data). The associated optimization problem, however, is of combinatorial nature and, hence, difficult to solve. In this work, we present an efficient implementation of a simple local search strategy that is based on matrix updates of the intermediate candidate solutions. Our experiments on both artificial and real-world data sets indicate that the approach can successfully incorporate unlabeled data in an efficient manner.
Downloadable publication This is an electronic reprint of the original article. |  |
