以關聯式法則建立共生式人工免疫系統之抗體集合

Abstract

在人工免疫系統中，抗體代表其問題解。而由基因演算法中合作式的概念衍伸出了共生式人工免疫系統，而抗體基元(partially specified antibody)在這當中則是代表其部分解。當抗體基元間非線性獨立而有相關聯性或者面對高維度的問題時，將抗體基元全部放在同一族群一起演化，容易使演算法效能低落。本論文提出以頻繁樣式成長法尋找抗體基元各維度間的關聯性，其關聯性的強弱將決定其分類規則。合理的分類規則可以使演算法在效能上有更佳的表現。因此對於共生式人工免疫系統來說，抗體基元分類為整個演化過程中相當重要的一部分。本論文由抗體基元分類根據維度關聯性訂立分類規則，可視問題不同做調整，在使用上可適用於多種問題類型。

In artificial immune system, antibodies represent the solutions. With the cooperative concept in genetic algorithm, the symbiotic artificial immune system has been developed, and partially specified antibodies are defined as the partial solutions. When the partially specified antibodies are not linear independent; have correlation with the others, or face the high-dimensional problem, put all partially specified antibodies in the same swarm which result in low efficiency in the algorithm. Frequent pattern growth algorithm is adopted in this thesis to find the correlation between each dimension, in which the strength of the connectivity determines whether the classification rules are valid or not. Generally, reasonable classification rules cause the algorithm work more effectively. Hence, for the algorithm, the partially specified antibody classification plays a very important role in the evolutionary process. The classification rules defined in this thesis are determined according to the correlation between each dimension, and they are adjustable under different problem, which greatly enhance the applicability of many kinds of problems.