適用於雲端運算環境之繼承式雙目標基因演算法應用與實作

Abstract

生物資訊學是利用資訊科學的方法研究生物學的問題。生物學家利用特徵擷取的方法，從大量複雜的數據中擷取出有用的資訊。其中，繼承式雙目標基因演算法被證明在解決選取最小量特徵集合與最大化正確率的最佳化問題上有很好的效率和準確率。然而，目前的繼承式雙目標基因演算法不具有可擴展性，無法善用平行運算機器的運算資源，且該應用無法普及讓所有生物學家都能使用。因此，本論文提出一套適用於雲端運算環境的優化方法並依據軟體即服務的模式建置一個容易使用的平台，達到可擴展性和易於取得的特性。 針對繼承式雙目標基因演算法中高運算量的部分，利用資料平行化和執行緒平行化的方式，使得該程式在多核心的機器上得到5.7倍的加速。此外，透過網頁介面，讓世界各地的生物學家能透過網際網路隨時取得運算資源，使用這項服務增進生物領域的研究，並藉此能提升繼承式雙目標基因演算法的影響力。

Bioinformatics is the application of information technology to the management of biological data in order to increase the understanding of biological processes. The biologists use the method of feature selection to obtain the informative data in the enormous and complex data. The inheritable bi-objective genetic algorithm has been proven to have more accuracy and efficiency to solve the optimization problem in selecting a minimal number of informative features while maximizing the prediction accuracy. However, the existing implementation of the inheritable bi-objective genetic algorithm may not be scalable when it runs on highly parallel machines and it is not widely accessible to all biologists. Therefore, this study aims to improve the scalability and accessibility of the application so that it could be used in a cloud computing environment using the SaaS (Software as a Service) model.

After effectively exploiting data parallelism and thread-level parallelism of some time consuming components in the inheritable bi-objective genetic algorithm application, the application can run 5.7 times faster than before. In addition, through the implementation of a web interface, the biologists with internet access could benefit from the application of this intelligent algorithm. This thesis work has increased the potential impact of the intelligent inheritable bi-objective genetic algorithm.