기계학습 군집을 위한 조화로운 메타휴리스틱 알고리즘

다국어 초록 (Multilingual Abstract)

Based on recent research, we need interaction between machine learning and optimization for big data analysis. Especially, we need to develop meta-heuristic clustering method because the number of clustering solutions is increasing exponentially with high number of clusters. We can find the global optimal clustering solution effectively considering diversified and converged search simultaneously for machine learning clustering. The objective of this research is to develop the harmonious meta-heuristic algorithm (HMHA) for clustering to control the diversified search in the initial stages and converged search in the final stages using initial solutions rate, population size, mutation rate and elitism. Our proposed HMHA is competitive comparing to previous methods based on our experiments and analysis.