Scholarlink

<p>The widespread emergence of Monkeypox has posed significant challenges to global public health. Early and accurate diagnosis is critical for effective management and containment. This study proposes a hybrid classification framework combining deep learning (DL) and supervised machine learning (ML) models to enhance monkeypox detection using skin lesion images. We extracted features using five pretrained Deep learning models—VGG19, DenseNet121, ResNet50, MobileNetV2, and InceptionV3 and integrated these features with various traditional and ensemble ML classifiers for classification purpose of Monkeypox disease. Our research includes 3 studies on two different datasets: the Monkeypox Skin Image Dataset (MSID) and Monkeypox Skin Lesion Detection (MSLD). The combination of the best suited methods can improve the overall accuracy of classification and the performance was evaluated using precision, recall, F1-score, and accuracy. According to our research, MobileNetV2 combined with KNN, CatBoost, and Extra Trees classifier yielded the best results on the MSID dataset (accuracy: 85.41%, 80.20%, and 79.16% respectively). For the MSLD dataset, DenseNet121 paired with KNN, CatBoost, and Random Forest achieved the highest accuracy (93.33%, 93.33%, and 95.56% respectively). The advantages and limitations of each integration are also discussed in this study.</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

Comparative analysis of supervised machine learning classifiers for classification of Monkeypox disease integrated with deep learning models

  • 期刊论文

  • 发表日期：

    2025年11月26日

查看全文
  • Nidhi Chauhan,
  • Sunita Yadav,
  • Alok Singh Chauhan

摘要

The widespread emergence of Monkeypox has posed significant challenges to global public health. Early and accurate diagnosis is critical for effective management and containment. This study proposes a hybrid classification framework combining deep learning (DL) and supervised machine learning (ML) models to enhance monkeypox detection using skin lesion images. We extracted features using five pretrained Deep learning models—VGG19, DenseNet121, ResNet50, MobileNetV2, and InceptionV3 and integrated these features with various traditional and ensemble ML classifiers for classification purpose of Monkeypox disease. Our research includes 3 studies on two different datasets: the Monkeypox Skin Image Dataset (MSID) and Monkeypox Skin Lesion Detection (MSLD). The combination of the best suited methods can improve the overall accuracy of classification and the performance was evaluated using precision, recall, F1-score, and accuracy. According to our research, MobileNetV2 combined with KNN, CatBoost, and Extra Trees classifier yielded the best results on the MSID dataset (accuracy: 85.41%, 80.20%, and 79.16% respectively). For the MSLD dataset, DenseNet121 paired with KNN, CatBoost, and Random Forest achieved the highest accuracy (93.33%, 93.33%, and 95.56% respectively). The advantages and limitations of each integration are also discussed in this study.