CAJ | 학술논문

遗传算法已经在许多领域得到了成功的应用，在解决模式识别问题时，由于遗传中的变异过程的存在导致了模式差异，如果能够减少这种差异性，就可以提高模式分类的正确率。针对这种情况，本文提出了在预分类基础上的一种逆变异算法，该算法采用遗传算法中变异过程的逆过程，来减少模式的差异性，达到提高模式分类正确率的目的。在ORL人脸库上的实验结果表明，该分类器的模式识别率明显高于传统的分类器算法及遗传算法。
유전산법이경재허다영역득도료성공적응용，재해결모식식별문제시，유우유전중적변이과정적존재도치료모식차이，여과능구감소저충차이성，취가이제고모식분류적정학솔。침대저충정황，본문제출료재예분류기출상적일충역변이산법，해산법채용유전산법중변이과정적역과정，래감소모식적차이성，체도제고모식분류정학솔적목적。재ORL인검고상적실험결과표명，해분류기적모식식별솔명현고우전통적분류기산법급유전산법。
Genetic Algorithms (GA) have been successfully applied in many important engineering areas. While solving pattern recognition problem, mutation process is the main cause of patterns differences in Genetic Algorithms. If reducing the differences, correct rate could be improved. According to this situation, this paper proposed a kind of Inverse Genetic Variant Algorithms on the basis of pre-classification. This Algorithm applied inverse process of varia-tion to reducing difference between patterns in order to prompting the correct rate. The experimental of ORL face data-base results indicated that the proposed classifier outperforms traditional methods in recognition accuracy.