A Comparative Study of $$\text {L}_1$$ L 1 and $$\text {L}_2$$ L 2 Norms in Support Vector Data Descriptions

Edgard M. Maboudou-Tchao () and Charles W. Harrison
Additional contact information
Edgard M. Maboudou-Tchao: University of Central Florida
Charles W. Harrison: University of Central Florida

A chapter in Control Charts and Machine Learning for Anomaly Detection in Manufacturing, 2022, pp 217-241 from Springer

Abstract: Abstract The Support Vector Data Description ( $$\text {L}_1$$ L 1 SVDD) is a non-parametric one-class classification algorithm that utilizes the $$\text {L}_1$$ L 1 norm in its objective function. An alternative formulation of SVDD, called $$\text {L}_2$$ L 2 SVDD, uses a $$\text {L}_2$$ L 2 norm in its objective function and has not been extensively studied. $$\text {L}_1$$ L 1 SVDD and $$\text {L}_2$$ L 2 SVDD are formulated as distinct quadratic programming (QP) problems and can be solved with a QP-solver. The $$\text {L}_2$$ L 2 SVDD and $$\text {L}_1$$ L 1 SVDD’s ability to detect small and large shifts in data generated from multivariate normal, multivariate t, and multivariate Laplace distributions is evaluated. Similar comparisons are made using real-world datasets taken from various applications including oncology, activity recognition, marine biology, and agriculture. In both the simulated and real-world examples, $$\text {L}_2$$ L 2 SVDD and $$\text {L}_1$$ L 1 SVDD perform similarly, though, in some cases, one outperforms the other. We propose an extension of the SMO algorithm for $$\text {L}_2$$ L 2 SVDD, and we compare the runtimes of four algorithms: $$\text {L}_2$$ L 2 SVDD (SMO), $$\text {L}_2$$ L 2 SVDD (QP), $$\text {L}_1$$ L 1 SVDD (SMO), and $$\text {L}_1$$ L 1 SVDD (QP). The runtimes favor $$\text {L}_1$$ L 1 SVDD (QP) versus $$\text {L}_2$$ L 2 SVDD (QP), sometimes substantially; however using SMO reduces the difference in runtimes considerably, making $$\text {L}_2$$ L 2 SVDD (SMO) feasible for practical applications. We also present gradient descent and stochastic gradient descent algorithms for linear versions of both the $$\text {L}_1$$ L 1 SVDD and $$\text {L}_2$$ L 2 SVDD. Examples using simulated and real-world data show that both methods perform similarly. Finally, we apply the $$\text {L}_1$$ L 1 SVDD and $$\text {L}_2$$ L 2 SVDD to a real-world dataset that involves monitoring machine failures in a manufacturing process.

Keywords: One-class classification; Support Vector Data Description; Sequential minimum optimization; L1-norm; L2-norm; SVDD; L2-SVDD; L1-SVDD; Quadratic programming; Gradient descent; Stochastic gradient descent; Monitoring; Manufacturing control chart; Machine failure (search for similar items in EconPapers)
Date: 2022
References: Add references at CitEc
Citations:

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:ssrchp:978-3-030-83819-5_9

Ordering information: This item can be ordered from
http://www.springer.com/9783030838195

DOI: 10.1007/978-3-030-83819-5_9

Access Statistics for this chapter

More chapters in Springer Series in Reliability Engineering from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().