本研究以螢火蟲最佳化(Glowworm Swarm Optimization, GSO)演算法為基礎,發展出新的資料分群方法,我們稱之為CGSO(Glowworm Swarm for Clustering)。GSO演算法是一種新的群體智慧技術,能夠求解連續型函數的最佳化問題,螢火蟲的亮度與解品質有關,其依賴亮度和半徑決定移動的方向,並會往較亮的螢火蟲移動,最終螢火蟲會聚集在多個位置上。本研究將資料分群問題設計為連續型最佳化問題並用GSO演算法求解。實驗結果顯示以GSO為基礎的分群演算法相較於K-Means和其他萬用啟發式演算法(Meta-Heuristic),如GA為基分群法、ACO為基分群法,能夠得到較佳的求解品值和穩定性,求解時間上也較ACO為基分群法SACO快速。
This paper presents a new data clustering algorithm based on glowworm swarm optimization (GSO) algorithm. GSO is a new type of swarm intelligence techniques and able to find solutions to optimization of continuous functions. In the proposed approach, data clustering problems are modeled as a continuous optimization problem and solved by using the GSO algorithm. The experimental results show that the GSO based clustering algorithm is very competitive compared to other meta-heuristic based approaches.