Scholarlink

Sigmoidal activation functions are frequently used in neural networks for introducing nonlinearity (Popa Electronics 12:24, 2023). Having a sigmoidal shape (S-shape), they have the important advantage of squishing input values into a limited range (usually between 0 and 1 or between − 1 and 1). It exists a multitude of sigmoidal activation functions, such as unipolar sigmoidal activation function, bipolar sigmoidal activation function, Einstein activation functions, hyperbolic tangent activation function or continuous-log sigmoidal activation functions.

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 Generating Sigmoidal Activation Functions

  • 章节

  • 发表日期：

    2025年

查看全文
  • Cosmin Radu Popa

摘要

Sigmoidal activation functions are frequently used in neural networks for introducing nonlinearity (Popa Electronics 12:24, 2023). Having a sigmoidal shape (S-shape), they have the important advantage of squishing input values into a limited range (usually between 0 and 1 or between − 1 and 1). It exists a multitude of sigmoidal activation functions, such as unipolar sigmoidal activation function, bipolar sigmoidal activation function, Einstein activation functions, hyperbolic tangent activation function or continuous-log sigmoidal activation functions.