Scholarlink

<p>Skin cancers are a major global public health problem that impacts people of all ages and demographics. This underscores the importance of advanced medical treatments and the development of sophisticated computer-aided diagnostic (CAD) systems to ensure precise classification and effective management of these conditions. With deep learning approaches and high-performance computation, computer vision tasks have advanced to the point that automatic classification of images related to skin diseases has become a popular field of study. Convolutional neural networks (CNNs) are often used in conjunction with ensembling techniques to further improve performance and increase the effectiveness of current approaches. However, traditional ensemble approaches, frequently result in an increased computing cost since they combine several baseline models. In contrast, the snapshot-based approaches, utilizing only a single baseline model, offer a streamlined alternative. In order to reduce the computational load associated with ensemble approaches, this paper presents a novel snapshot-based feature fusion approach coupled with a machine learning-based classifier, SEFFNet, which allows the capturing of more diverse feature representations at different training stages without the need for multiple models, thereby mitigating the computational burden associated with ensemble approaches. Additionally, our approach refrains from employing data augmentation and pre-processing techniques to mitigate computational overload. To demonstrate the effectiveness of the proposed method, we conduct experiments on a variety of publicly accessible datasets, including the HAM10000 and the ISIC 2017 datasets. The suggested approach achieves an accuracy of 92.93% on the HAM10000 and 81.17% on the ISIC 2017 datasets. The source code is available at: <a href="https://github.com/Cmatermedicalimageanalysis/SEFFNet">https://github.com/Cmatermedicalimageanalysis/SEFFNet</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

SEFFNet: snapshot ensemble-based feature fusion network for skin cancer classification

  • 期刊论文

  • 发表日期：

    2026年2月22日

查看全文
  • Soumyajit Gayen,
  • Sourajit Maity,
  • Sujan Sarkar,
  • Erik Cuevas,
  • Ram Sarkar

摘要

Skin cancers are a major global public health problem that impacts people of all ages and demographics. This underscores the importance of advanced medical treatments and the development of sophisticated computer-aided diagnostic (CAD) systems to ensure precise classification and effective management of these conditions. With deep learning approaches and high-performance computation, computer vision tasks have advanced to the point that automatic classification of images related to skin diseases has become a popular field of study. Convolutional neural networks (CNNs) are often used in conjunction with ensembling techniques to further improve performance and increase the effectiveness of current approaches. However, traditional ensemble approaches, frequently result in an increased computing cost since they combine several baseline models. In contrast, the snapshot-based approaches, utilizing only a single baseline model, offer a streamlined alternative. In order to reduce the computational load associated with ensemble approaches, this paper presents a novel snapshot-based feature fusion approach coupled with a machine learning-based classifier, SEFFNet, which allows the capturing of more diverse feature representations at different training stages without the need for multiple models, thereby mitigating the computational burden associated with ensemble approaches. Additionally, our approach refrains from employing data augmentation and pre-processing techniques to mitigate computational overload. To demonstrate the effectiveness of the proposed method, we conduct experiments on a variety of publicly accessible datasets, including the HAM10000 and the ISIC 2017 datasets. The suggested approach achieves an accuracy of 92.93% on the HAM10000 and 81.17% on the ISIC 2017 datasets. The source code is available at: https://github.com/Cmatermedicalimageanalysis/SEFFNet.