First published 2 September 2013
Accelerating Immunos 99
Paul Taylor, Fiona Polack, Jon Timmis
Immunos 99 is a classification algorithm based upon the principles of Artificial Immune Systems (AIS). AIS algorithms can provide alternatives to classical techniques such as decision trees for classification tasks. Immunos 99 provides one alternative however the algorithm and implementations have some room for performance improvement. This paper discusses improvements made to Immunos 99 and its reference implementation to improve runtime performance. The new algorithm/implementation results in an approximate 40% reduction in run time on test data. The paper closes with a proposal for an implementation of the Immunos 99 algorithm that is suitable for use on map/reduce clusters.