Scholarlink

In addition to the functions from previous two chapters (sigmoidal activation functions, presented in Chap. 1 and radial basis activation functions described in Chap. 2 ), it exists other important mathematical functions, having a multitude of applications in analog signal processing and artificial neural networks (Popa EURASIP J Adv Sig Process 2012:129, 2012; Popa Electronics 12:24, 2023). These functions are detailed analyzed in this chapter and the most accurate approximation functions are proposed, in order to generate them. Two important objectives are considered: as good as possible accuracy of the approximation and reasonable hardware resources required for their implementation in CMOS technology using fundamental CMOS computational circuits (further described in Chap. 4 ).

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Superior-Order Approximation Functions for Artificial Neural Networks Applications

  • 章节

  • 发表日期：

    2025年

查看全文
  • Cosmin Radu Popa

摘要

In addition to the functions from previous two chapters (sigmoidal activation functions, presented in Chap. 1 and radial basis activation functions described in Chap. 2 ), it exists other important mathematical functions, having a multitude of applications in analog signal processing and artificial neural networks (Popa EURASIP J Adv Sig Process 2012:129, 2012; Popa Electronics 12:24, 2023). These functions are detailed analyzed in this chapter and the most accurate approximation functions are proposed, in order to generate them. Two important objectives are considered: as good as possible accuracy of the approximation and reasonable hardware resources required for their implementation in CMOS technology using fundamental CMOS computational circuits (further described in Chap. 4 ).