Using the genetic algorithm to build optimal neural networks for fault-prone module detection

The genetic algorithm is applied to developing optimal or near optimal backpropagation neural networks for fault{prone/not{fault{prone classiication of software modules. The algorithm considers each network in a population of neural networks as a potential solution to the optimal classiication problem. Variables governing the learning and other parameters and network architecture are represented as substrings (genes) in a machine{level bit string (chromosome). When the population undergoes simulated evolution using genetic operators | selection based on a tness function, crossover, and mutation | the average performance increases in successive generations. We found that, on the same data, compared with the best manually developed networks, evolved networks produced improved classiications in considerably less time, with no human eeort, and with greater conndence in their optimality or near optimality. Strategies for devising a tness function speciic to the problem are explored and discussed .