Skip to content
Open access

A novel optimized deep ensemble for monkeypox lesion detection using transfer learning and ABC algorithm

Aug 2026 · Discover Computing · Vol 29 · 0 citations · 54 references

TL;DR

This paper proposes MPox-Net1.0: An Artificial Bee Colony (ABC)-optimized weighted ensemble of pre-trained convolutional neural networks (CNNs) for monkeypox classification that integrates transfer learning with swarm intelligence to dynamically balance model contributions based on prediction confidence.

Abstract

Monkeypox, a reemerging zoonotic disease, has become a serious global health concern that requires rapid and accurate diagnostic methods. The increasing prevalence of skin lesion images highlights the scope of automated deep learning solutions. However, current approaches face challenges such as limited generalization and poor performance on imbalanced datasets. This paper aims to develop a robust diagnostic framework by addressing these limitations through a novel ensemble learning strategy. We propose MPox-Net1.0: An Artificial Bee Colony (ABC)-optimized weighted ensemble of pre-trained convolutional neural networks (CNNs) for monkeypox classification. The methodology integrates transfer learning with swarm intelligence to dynamically balance model contributions based on prediction confidence. Eight state-of-the-art CNNs were fine-tuned on the MSLD v2.0 dataset, with Xception, EfficientNetB3, and EfficientNetB0 selected for the ensemble based on performance. The proposed ABC-weighted ensemble achieved an accuracy of 99.12%, improved by 0.66% over the best individual model. It also achieved better precision (99.15%), recall (99.12%), and F1 score (99.12%) compared to existing ensemble methods. Overall, our technique demonstrates enhanced minority class recognition and computational efficiency, establishing its potential for clinical deployment.

Read PDF

Similar papers

Conference Open access Sep 2026

Skin Lesion Image Classification Based on Deep Learning: A Systematic Ablation Study of Heterogeneous CNN Ensembles

A heterogeneous CNN voting ensemble comprising ResNet50, ResNet101, and EfficientNet -B0 for 7-class skin lesion classification on the ISIC2018 dataset is constructed, establishing a reusable analytical framework for rigorous ensemble evaluation in broader medical imaging classification tasks.

Jin-Fu Ye · 0 citations
Open access Aug 2026

An Explainable AI Ensemble Model for Acquired Vitelliform Lesion Identification

Abstract Retinal diseases gradually weaken eyesight and may even potentially lead to blindness. Recognizing changes in the retina based on OCT imaging allows for the detection of diseases at their early stages and thus for making an appropriate diagnosis. In this study, acquired vitelliform lesions (AVL), drusen, and h...

Paweł Powroźnik, M. Skublewska-Paszkowska, K. Nowomiejska et al. · 0 citations
Open access Aug 2026

Enhancing Interpretability for Automated Human Monkeypox Skin Lesions Detection Using Pre-trained Transfer Learning Networks

MobileNetV3 provides the most balanced trade-off among the evaluated models between classification accuracy, robustness, and computational efficiency, and these evaluated performances indicate the potential of the HMSLD framework.

Kanak Kumar, Anshul Verma, Pradeepika Verma · 0 citations
Review Open access Aug 2026

A trustworthy cross-domain AI framework for fundus disease classification using hybrid CNN fusion and supervised domain adaptation

FusionEye-Net combines strong internal performance with substantially improved cross-domain generalization through lightweight adaptation, underscoring the critical role of external validation and domain-aware fine-tuning in formulating trustworthy, adaptive AI-assisted decision-support tools that can safely augment au...

Ali M. Duhaim, A. M. Al-Bakry · 0 citations
Open access Oct 2026

A Dual-Stream Feature Learning Framework with Adaptive Bio-Inspired Protozoa–Starfish Optimization for Melanoma Classification

Melanoma accounts for a small fraction of skin cancer cases but for the majority of skin-cancer-related deaths, making accurate early detection essential. Although deep learning-based dermoscopic image analysis has shown strong diagnostic potential, its performance depends strongly on training hyperparameters and loss-...

Sreelakshmi Jayasankar, T. Brindha · 0 citations
Sep 2026

HybridDermNet: A CNN–Transformer Framework with Latent Bottleneck Learning for Robust Skin Cancer Classification

A CNN–Transformer-based framework with latent bottleneck learning for robust multi-class skin cancer classification is proposed, which offers compact, interpretable and generalizable representations for reliable dermatology decision support in heterogeneous imaging conditions.

Vunnam Narmada, Kalavala Asish Vardhan · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.