KACU: k-means with hardware centroid-updating

In this paper, we propose a framework, KACU (standing for k-means with hardware centroid updating), to enhance the speed of k-means clustering algorithm by integrating a hardware centroid updating mechanism into the procedure of continuous k-means algorithm. To facilitate performance measurement, KACU is implemented in a commercial field programmable gate array (abbreviated as FPGA) device. The experimental results show that KACU is able to achieve considerably higher performance.