TY - CHAP

T1 - Active and semi-supervised data domain description

AU - Görnitz, Nico

AU - Kloft, Marius

AU - Brefeld, Ulf

PY - 2009/7/1

Y1 - 2009/7/1

N2 - Data domain description techniques aim at deriving concise descriptions of objects belonging to a category of interest. For instance, the support vector domain description (SVDD) learns a hypersphere enclosing the bulk of provided unlabeled data such that points lying outside of the ball are considered anomalous. However, relevant information such as expert and background knowledge remain unused in the unsupervised setting. In this paper, we rephrase data domain description as a semi-supervised learning task, that is, we propose a semi-supervised generalization of data domain description (SSSVDD) to process unlabeled and labeled examples. The corresponding optimization problem is non-convex. We translate it into an unconstraint, continuous problem that can be optimized accurately by gradient-based techniques. Furthermore, we devise an effective active learning strategy to query low-confidence observations. Our empirical evaluation on network intrusion detection and object recognition tasks shows that our SSSVDDs consistently outperform baseline methods in relevant learning settings.

AB - Data domain description techniques aim at deriving concise descriptions of objects belonging to a category of interest. For instance, the support vector domain description (SVDD) learns a hypersphere enclosing the bulk of provided unlabeled data such that points lying outside of the ball are considered anomalous. However, relevant information such as expert and background knowledge remain unused in the unsupervised setting. In this paper, we rephrase data domain description as a semi-supervised learning task, that is, we propose a semi-supervised generalization of data domain description (SSSVDD) to process unlabeled and labeled examples. The corresponding optimization problem is non-convex. We translate it into an unconstraint, continuous problem that can be optimized accurately by gradient-based techniques. Furthermore, we devise an effective active learning strategy to query low-confidence observations. Our empirical evaluation on network intrusion detection and object recognition tasks shows that our SSSVDDs consistently outperform baseline methods in relevant learning settings.

KW - Informatics

KW - Active Learning

KW - Background knowledge

KW - Baseline methods

KW - Continuous problems

KW - Data domain description

KW - Empirical evaluations

KW - Gradient based

KW - Learning settings

KW - Network intrusion detection

KW - Optimization problems

KW - Semi-supervised learning

KW - upport vector domain description

KW - Unlabeled data

KW - Business informatics

UR - http://www.scopus.com/inward/record.url?scp=70350627210&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-04180-8_44

DO - 10.1007/978-3-642-04180-8_44

M3 - Article in conference proceedings

AN - SCOPUS:70350627210

SN - 978-3-642-04179-2

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 407

EP - 422

BT - Machine Learning and Knowledge Discovery in Databases

A2 - Buntine, Wray

A2 - Grobelnik, Marko

A2 - Mladenic, Dunja

A2 - Shawe-Taylor, John

PB - Springer

CY - Berlin, Heidelberg

T2 - European Conference on Machine Learning and Knowledge Discovery in Databases - 2009

Y2 - 7 September 2009 through 11 September 2009

ER -