Scholarlink

This paper explores the development of advanced neural network architectures for image denoising. By integrating fractional calculus into deep convolutional neural networks, we introduce an optimal control framework, where network weights are iteratively optimized using Primal-Dual algorithm. This approach yields a highly efficient and faster numerical algorithm for image denoising. To validate the method’s effectiveness, we showcase experimental results demonstrating superior performance compared to existing techniques.

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

A Primal-Dual Approach to Fractional Optimal Control in Non-smooth Machine Learning: FODE-Based Solutions for Enhanced Image Denoising

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Fakhr-eddine Limami,
  • Amine Laghrib,
  • Lekbir Afraites

摘要

This paper explores the development of advanced neural network architectures for image denoising. By integrating fractional calculus into deep convolutional neural networks, we introduce an optimal control framework, where network weights are iteratively optimized using Primal-Dual algorithm. This approach yields a highly efficient and faster numerical algorithm for image denoising. To validate the method’s effectiveness, we showcase experimental results demonstrating superior performance compared to existing techniques.