Profiling novel classification algorithms Artificial Immune Systems

In this paper we present an approach for bench-marking and profiling novel classification algorithms. We apply it to AIRS, an artificial immune system algorithm inspired by how the natural immune system recognizes and remembers intruders. We provide basic benchmarking results for AIRS, to our knowledge the first such test under standardised conditions. We also investigate how data set properties (data set size) relate to AIRS performance, and what other algorithms produce similar patterns over over- and underperformance on specific data sets. We present three methods for computing algorithm similarity that may be useful for profiling novel algorithms in general.