CAJ | 학술논문

针对危险状态识别问题,提出一种改进的树突状细胞算法(IDCA).在原算法的基础上引入"时间窗"、未成熟DC尽快成熟、衰减因子等概念与策略,提高算法的响应速度和识别精度,解决传统树突状细胞算法一遍运行可能无法评价输入序列末端抗原的问题.在UCI标准数据集上的对比实验证明了IDCA算法的有效性.
침대위험상태식별문제,제출일충개진적수돌상세포산법(IDCA).재원산법적기출상인입"시간창"、미성숙DC진쾌성숙、쇠감인자등개념여책략,제고산법적향응속도화식별정도,해결전통수돌상세포산법일편운행가능무법평개수입서렬말단항원적문제.재UCI표준수거집상적대비실험증명료IDCA산법적유효성.
Aiming at the detect the danger status, this paper proposes an Improved Dendritic Cell Algorithm(IDCA). It adopts the idea of "time window", takes the strategy of maturing as soon as possible to the immature DCs, and introduces the attenuation factor based on the original algorithm to achieve the advantages of short response time and higher recognition precision. The problem that DCA may not be able to evaluate the antigens located at the end of input time sequence if only running algorithm once is also solved. The experiments on UCI dataset prove that IDCA is really available.