Visible to the public Hardware Design of Gaussian Kernel Function for Non-Linear SVM Classification

TitleHardware Design of Gaussian Kernel Function for Non-Linear SVM Classification
Publication TypeConference Paper
Year of Publication2021
AuthorsWang, Yuanfa, Pang, Yu, Huang, Huan, Zhou, Qianneng, Luo, Jiasai
Conference Name2021 IEEE 14th International Conference on ASIC (ASICON)
KeywordsClassification algorithms, Conferences, digital computers, exponentiation, Hardware, Multiplexing, pubcrawl, resilience, Resiliency, Resource management, Scalability, Support vector machines
AbstractHigh-performance implementation of non-linear support vector machine (SVM) function is important in many applications. This paper develops a hardware design of Gaussian kernel function with high-performance since it is one of the most modules in non-linear SVM. The designed Gaussian kernel function consists of Norm unit and exponentiation function unit. The Norm unit uses fewer subtractors and multiplexers. The exponentiation function unit performs modified coordinate rotation digital computer algorithm with wide range of convergence and high accuracy. The presented circuit is implemented on a Xilinx field-programmable gate array platform. The experimental results demonstrate that the designed circuit achieves low resource utilization and high efficiency with relative error 0.0001.
DOI10.1109/ASICON52560.2021.9620361
Citation Keywang_hardware_2021