K-means + + centroid initialization-based k-means algorithm hardware implementation method and system

• Main Authors: ZHOU YUCHEN, FU YUXIANG, HE GUOQIANG, LI WEI, LI LI, HE SHUZHUAN
• Format: Patent
• Language: Chinese; English
• Published: 17.04.2020
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention relates to a k-means + + centroid initialization-based k-means algorithm hardware implementation method and system, and the system comprises: a main control module which provides controland reconstruction information for algorithm implementation; a storage control module which is used for controlling data transmission and storage; a centroid initialization module which is used for initializing a centroid based on a k-means + + algorithm; a clustering operation module which is used for obtaining a clustering result based on distance comparison; a centroid updating module which isused for calculating and updating the centroid through the average value of each category; and a threshold comparison module which is used for judging whether the centroid converges or not. Accordingto the hardware implementation method, the parallelism of hardware is fully utilized, a k-means + + centroid initialization scheme is used for replacing a random centroid initialization scheme, and the algorithm convergence sp