Fig. 2

Statistical GA algorithm workflow after including the use of topic models. This figure describes the updated flow of the Statistical GA algorithm after adding the perplexity measurement for the selection of solutions. There are four steps in this flow: the first one, in which the candidate instances are selected by the original Statistical GA; the second one, in which these instances are clustered attending to their similarity calculated by their hamming distance; the third one, which consists of building the CTM and measuring its perplexity, and the last step, which consists of reporting the motif if the perplexity calculated in the previous step is lower than 100