Clique finding-a genetic approach

Presents a novel and efficient genetic approach for finding maximal cliques in a graph. A binary alphabet has been chosen to represent the presence or absence of nodes in a subgraph. The approach is to start with an initial population having small sized graphs, and then to effectively generate larger ones using a new crossover mechanism called `partial copy crossover´. The splitting of the mutation operator into two operators, namely `addition´ and `deletion´, has been found to be effective for both increasing the diversity of the population and controlling the number of relevant subgraphs, i.e. those with the potentiality to become cliques. Experimental results on graphs with between 5 and 50 nodes and varying edge densities establish the efficacy and robustness of the approach