A GPU-accelerated Density-Based Clustering Algorithm
Woong-Kee Loh, Young‐Kuk Kim · 2014
Due to the advances in GPU technology, there have been many approaches to utilize the GPU for general applications. Many research papers that dramatically improved the performance of traditional CPU-based data mining algorithms have been published. Clustering is an important data mining problem that is often found in many areas. DBSCAN is the most widely used density-based clustering algorithm, but it has a drawback that the optimal parameters can be hardly found. OPTICS was proposed to tackle the problem. In this paper, we propose an algorithm that significantly improves the performance of OPTICS using the GPU. Through extensive experiments, we show that our algorithm outperforms OPTICS by an order of magnitude.