Scholarlink

<p>We introduce a novel computational unit for neural networks that features multiple biases, challenging the traditional perceptron structure. This unit emphasizes the importance of preserving uncorrupted information as it is passed from one unit to the next, applying activation functions later in the process with specialized biases for each unit. Through both empirical and theoretical analyses, we show that by focusing on increasing biases rather than weights, there is potential for significant enhancement in a neural network model’s performance. This approach offers an alternative perspective on optimizing information flow within neural networks. See source code (CurioSAI in Increasing biases can be more efficient than increasing weights, 2023. <a href="https://github.com/CuriosAI/dac-dev">https://github.com/CuriosAI/dac-dev</a>).</p>

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

Increasing biases can be more efficient than increasing weights

  • 期刊论文

  • 发表日期：

    2025年6月16日

查看全文
  • Carlo Metta,
  • Marco Fantozzi,
  • Andrea Papini,
  • Gianluca Amato,
  • Matteo Bergamaschi,
  • Andrea Fois,
  • Silvia Giulia Galfrè,
  • Alessandro Marchetti,
  • Michelangelo Vegliò,
  • Maurizio Parton,
  • Francesco Morandin

摘要

We introduce a novel computational unit for neural networks that features multiple biases, challenging the traditional perceptron structure. This unit emphasizes the importance of preserving uncorrupted information as it is passed from one unit to the next, applying activation functions later in the process with specialized biases for each unit. Through both empirical and theoretical analyses, we show that by focusing on increasing biases rather than weights, there is potential for significant enhancement in a neural network model’s performance. This approach offers an alternative perspective on optimizing information flow within neural networks. See source code (CurioSAI in Increasing biases can be more efficient than increasing weights, 2023. https://github.com/CuriosAI/dac-dev).