Genetic Optimization Algorithm And Its Application In Data Mining

Mathematic base, searching mechanism and features of genetic algorithm have been analyzed completely. And in order to overcome early convergence and improve effectiveness of searching and exploring, following work has been done:1. In order to prevent algorithm from entering to local max, disturbing implement strategy has been brought forward, that is put a disturb on new individuals and then judge whether to accept it, so bad individuals can't be accepted unconditionally;2. In order to help algorithm jump out of local max, simulating ant searching food strategy has been brought forward. It is designed based on ant's searching food behavior. Its core is when algorithm enters into local max,, put individuals to cross with idle individuals and replace those by spared individuals which don't have cross with idle individuals.3. In order to improve searching quality of restricting optimization, good and bad gene code method has been brought forward. Its focus lays in emphasizing the value of gene to targets and more precisely simulate evolvement mechanism so that useful information in problem space will be used completely.4. Application of genetic algorithm in data mining has been studied. Based on gene hypothesis gene dynamic sort method has been brought forward. Its feature is component of genes can become longer and better gene component and because of dynamic sort, good gene component can be kept effectively.