Scholarlink

<p>Spiking neural networks (SNNs) and spiking neural P systems (SN P Systems), inspired by biological behaviors, are computational models that offer significant advantages, particularly in energy-efficient computation and temporal information processing. Due to the discrete nature of spikes, these networks cannot directly utilize common learning mechanisms like backpropagation. Hence, using the foundational computational frameworks, this narrative review article explores the various learning mechanisms. For SNNs, it includes supervised, unsupervised, and reinforcement learning. Additionally, the review provides methods for converting ANNs into spike-based models. In parallel, it provides the integration of learning strategies into SN P systems. The working principles of algorithms for the learning mechanisms have been included. Also, due to the energy-efficient nature of these models, they are also used in a diverse range of applications, and this review explores recent applications where these learning methods are used. It provides and highlights various simulation frameworks for the efficient design and deployment of these models. Finally, the discussion extends to the comparison and strengths, outlines current challenges and future directions for research, and expands the capabilities of these models in real-world applications.</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

Computational frameworks, learning, and applications of spiking-based models of computation: a narrative review

  • 期刊论文

  • 发表日期：

    2026年3月23日

查看全文
  • Sushant Yadav,
  • Naveen Gehlot,
  • Santosh Chaudhary,
  • Rajesh Kumar,
  • Pilani Nkomozepi

摘要

Spiking neural networks (SNNs) and spiking neural P systems (SN P Systems), inspired by biological behaviors, are computational models that offer significant advantages, particularly in energy-efficient computation and temporal information processing. Due to the discrete nature of spikes, these networks cannot directly utilize common learning mechanisms like backpropagation. Hence, using the foundational computational frameworks, this narrative review article explores the various learning mechanisms. For SNNs, it includes supervised, unsupervised, and reinforcement learning. Additionally, the review provides methods for converting ANNs into spike-based models. In parallel, it provides the integration of learning strategies into SN P systems. The working principles of algorithms for the learning mechanisms have been included. Also, due to the energy-efficient nature of these models, they are also used in a diverse range of applications, and this review explores recent applications where these learning methods are used. It provides and highlights various simulation frameworks for the efficient design and deployment of these models. Finally, the discussion extends to the comparison and strengths, outlines current challenges and future directions for research, and expands the capabilities of these models in real-world applications.