Scholarlink

<p>Conventional methods for deep clustering often suffer from slow convergence and issues related to misassignments, which can compromise confidentiality. This paper addresses these problems by introducing a novel Deep Fuzzy Clustering Neural Network (DFC-NN). Our approach leverages deep learning techniques to dynamically refine representations and clusters simultaneously through an optimization process. Unlike traditional methods that require manual updates, DFC-NN adjusts centroids using learned gradients, ensuring adaptability to evolving data distributions. This integrated approach not only enhances clustering precision and comprehensiveness, but also streamlines the overall process. We validate the robustness and effectiveness of DFC-NN through extensive experiments on benchmark datasets such as MNIST, USPS, F-MNIST, and CIFAR-10, demonstrating that our model outperforms state-of-the-art methods in clustering performance. The implementation of DFC-NN will be made publicly available at: <a href="https://github.com/ChaimaDerouiche/Deep-Fuzzy-Clustering-Neural-Network-DFC-NN-Fast-Convergence-and-Enhanced-Clustering-Performance">https://github.com/ChaimaDerouiche/Deep-Fuzzy-Clustering-Neural-Network-DFC-NN-Fast-Convergence-and-Enhanced-Clustering-Performance</a> upon publication.</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

Deep fuzzy clustering neural network (DFC-NN): fast convergence and enhanced clustering performance

  • 期刊论文

  • 发表日期：

    2025年9月11日

查看全文
  • Chaima Derouiche,
  • Abdallah Bensayah,
  • Khadra Bouanane,
  • Oussama Aiadi,
  • Belal Khaldi,
  • Mohammed Lamine Kherfi

摘要

Conventional methods for deep clustering often suffer from slow convergence and issues related to misassignments, which can compromise confidentiality. This paper addresses these problems by introducing a novel Deep Fuzzy Clustering Neural Network (DFC-NN). Our approach leverages deep learning techniques to dynamically refine representations and clusters simultaneously through an optimization process. Unlike traditional methods that require manual updates, DFC-NN adjusts centroids using learned gradients, ensuring adaptability to evolving data distributions. This integrated approach not only enhances clustering precision and comprehensiveness, but also streamlines the overall process. We validate the robustness and effectiveness of DFC-NN through extensive experiments on benchmark datasets such as MNIST, USPS, F-MNIST, and CIFAR-10, demonstrating that our model outperforms state-of-the-art methods in clustering performance. The implementation of DFC-NN will be made publicly available at: https://github.com/ChaimaDerouiche/Deep-Fuzzy-Clustering-Neural-Network-DFC-NN-Fast-Convergence-and-Enhanced-Clustering-Performance upon publication.