Using a spectral technique, genetic algorithms and decision diagrams for finding unconditional table tests

Karen O. Egiazarian, Radomir S. Stankovic, Jaakko Astola, Heikki Huttunen

Abstract: In this paper, we study the problem of finding minimal unconditional tests for a given table. The earlier spectral method of solving this problem is modified into a form using decision diagrams and a form using genetic algorithms. The decision diagram method extends the spectral algorithm for larger tables. The genetic algorithm can be used to find a near-optimal solution for very large tables that are intractable with the other methods.