Augmented negative selection algorithm with variable-coverage detectors

Abstract

An augmentation of negative selection algorithm is developed featuring detectors that have variable coverage. While the detectors can have different kinds of variable properties in the light of this concept, the paper mainly describes the experiments of variable-sized detectors in real-valued space. Effects of the two main control parameters, self radius and expected coverage, are discussed and experimented with both synthesized and real-word datasets. The new approach improves efficiency and reliability without compromising the order of magnitude of complexity.

Publication Title

Proceedings of the 2004 Congress on Evolutionary Computation, CEC2004

This document is currently not available here.

Share

COinS