在可程式化系統晶片上之C-Means分群演算法設計

Abstract

本論文提出一個新的c-means演算法硬體架構，在做分群處理與質量中心點的計算皆為管線化的運作，因此可以同時處理多筆訓練向量。我們計算質量中心點的除法器是由查表法、乘法器與位移運作所組成，可以大大的降低硬體複雜度且一個時脈週期即可完成除法的運算。最後我們所提出的架構會在以FPGA為基礎的可程式化系統晶片設計(System On a Programmable Chip，SOPC)之平台上作實際的效能測試，經由數據分析可以發現我們的架構會比軟體有更高的效能。A novel hardware architecture for c-means clustering is presented in this paper. Our architecture is fully pipelined for both the partitioning and centroid computation operations so that multiple training vectors can be concurrently processed. A simple divider circuit based on table lookup, multiplication and shift operations is employed for reducing both the area cost and latency for centroid computation. The proposed architecture is used as a hardware accelerator for a softcore NIOS CPU implemented on a FPGA device for physical performance measurement.Numerical results reveal that our design is an effective solution with low hardware complexity and high computation performance for c-means design.