本研究針對布穀鳥搜尋演算法(Cuckoo Search, CS)及Lévy flight的相關機制來進行探討。為了近一步地提高布穀鳥搜尋演算法的效率,以它為框架進行改良,來提高搜尋的能力。因此本研究參考了許多文獻,提出了改良型動態適應布穀鳥搜尋演算法(Improved Dynamic Adaptation Cuckoo Search Algorithm, IDACS)。為了驗證本研究所提出的IDACS的搜尋能力,針對了兩種實驗進行探討。第一,對數學型的測試函數來實驗;第二,對UCI資料庫的真實數據集做類神經網路訓練實驗。最後由實驗結果可以證明,在數學型的測試函數中,IDACS達到成功門檻的比例皆為100%,明顯的比CS的搜尋能力還要好。在真實環境應用下,確實也比較出色,且執行的時間減少約40%,收斂速度更快。
This study for Cuckoo Search (CS) and Lévy flight mechanisms were discussed. In order to further improve the efficiency of Cuckoo Search, use it as a framework for improved to enhance search capabilities. Therefore, this study refers to a lot of literature, presented the Improved Dynamic Adaptation Cuckoo Search Algorithm (IDACS). In order to validate the presented IDACS search capabilities, two experiments were discussed. First, the function experiment. Second, for the UCI database of real experimental data did Artificial Neural Network training. Finally the result proved that in function, the IDACS achieve success threshold proportion were a hundred percent, obviously better than CS search capabilities. In real environment applications, indeed better, and reduced about forty percent the time, faster convergence.