Is Negative Selection Appropriate for Anomaly Detection ?
Negative selection algorithms for hamming and realvalued shapespaces are reviewed. Problems are identified with the use of these shapespaces, and the negative selection algorithm in general, when applied to anomaly detection. A straightforward self detector classification principle is proposed and its classification performance is compared to a realvalued negative selection algorithm and to a oneclass support vector machine. Earlier work suggests that realvalue negative selection requires a single class to learn from. The investigations presented in this paper reveal, however, that when applied to anomaly detection, the realvalued negative selection and self detector classification techniques require positive and negative examples to achieve a high classification accuracy. Whereas, oneclass SVMs only require examples from a single class.
Is Negative Selection Appropriate for Anomaly Detection ?
Proceedings of the ACM SIGEVO Genetic and Evolutionary Computation Conference (GECCO2005)
Authors:  Thomas Stibor, P. Mohr, J. Timis, and Claudia Eckert 
Year/month:  2005/ 
Booktitle:  Proceedings of the ACM SIGEVO Genetic and Evolutionary Computation Conference (GECCO2005) 
Address:  Washington, D.C. 
Publisher:  ACM Press 
Fulltext: 
Abstract 

Negative selection algorithms for hamming and realvalued shapespaces are reviewed. Problems are identified with the use of these shapespaces, and the negative selection algorithm in general, when applied to anomaly detection. A straightforward self detector classification principle is proposed and its classification performance is compared to a realvalued negative selection algorithm and to a oneclass support vector machine. Earlier work suggests that realvalue negative selection requires a single class to learn from. The investigations presented in this paper reveal, however, that when applied to anomaly detection, the realvalued negative selection and self detector classification techniques require positive and negative examples to achieve a high classification accuracy. Whereas, oneclass SVMs only require examples from a single class. 
Bibtex:
@inproceedings { Stibor2005b,author = { Thomas Stibor and P. Mohr and J. Timis and Claudia Eckert},
title = { Is Negative Selection Appropriate for Anomaly Detection ? },
year = { 2005 },
booktitle = { Proceedings of the ACM SIGEVO Genetic and Evolutionary Computation Conference (GECCO2005) },
address = { Washington, D.C. },
publisher = { ACM Press },
}