Application of EfficientNetV2-S Architecture with Focal Loss to Overcome Class Imbalances in Skin Cancer Classification

Marfungah  Wati; Aris  Thobirin; Sugiyarto  Surono

doi:10.59395/ijadis.v7i1.1524

Authors

Marfungah Wati Department of Mathematics, Ahmad Dahlan University, Indonesia
Aris Thobirin Department of Mathematics, Ahmad Dahlan University, Indonesia
Sugiyarto Surono Department of Mathematics, Ahmad Dahlan University, Indonesia

DOI:

https://doi.org/10.59395/ijadis.v7i1.1524

Keywords:

EfficientNetV2-S, Focal Loss, Class Imbalance, Skin Cancer, Dermoscopic Images

Abstract

Imbalanced class distributions in skin lesion image datasets can reduce the effectiveness of multiclass classification models. This research proposes a classification model based on the EfficientNetV2-S architecture with the application of two-stage training and loss functions that emphasize learning in classes with limited data. The models were trained using on-the-fly image augmentation and evaluated to assess generalization capabilities to the test data. In the initial stage, the model is trained by freezing the backbone and only updating the classifier layer. Next, fine-tuning was carried out on part of the backbone layer to adjust the representation of features to the image characteristics of the skin lesion. Evaluation is conducted through multiple training times with different random initializations to ensure consistency of results. The test results showed that the model experienced an improvement in performance after the fine-tuning process, with an accuracy of about 88% as well as an increase in F1-score values in some classes. Overall, the results indicate that the proposed approach may help improve classification performance when dealing with imbalanced skin cancer image data.

1100 519

Downloads

Download data is not yet available.

References

[1] A. Aldwean and D. Tenney, Artificial Intelligence in Healthcare Sector: A Literature Review of the Adoption Challenges, pp. 129147, 2024, doi: 10.4236/ojbm.2024.121009. DOI: https://doi.org/10.4236/ojbm.2024.121009

[2] L. Choudhary and J. S. Choudhary, Deep Learning Meets Machine Learning: A Synergistic Approach towards Artificial Intelligence, J. Sci. Res. Reports, vol. 30, no. 11, pp. 865875, 2024, doi: 10.9734/jsrr/2024/v30i112614. DOI: https://doi.org/10.9734/jsrr/2024/v30i112614

[3] J. A. Alhijaj and R. S. Khudeyer, Techniques and Applications for Deep Learning: A Review, J. Al-Qadisiyah Comput. Sci. Math., vol. 15, no. 2, pp. 112, 2023, doi: 10.29304/jqcm.2023.15.2.1236. DOI: https://doi.org/10.29304/jqcm.2023.15.2.1236

[4] K. Aggarwal et al., Has the Future Started? The Current Growth of Artificial Intelligence, Machine Learning, and Deep Learning, Iraqi J. Comput. Sci. Math., vol. 3, no. 1, pp. 115123, 2022, doi: 10.52866/ijcsm.2022.01.01.013. DOI: https://doi.org/10.52866/ijcsm.2022.01.01.013

[5] F. Badri, M. Taqijuddin Alawiy, and E. M. Yuniarno, Deep Learning Architecture Based on Convolutional Neural Network (Cnn) on Animal Image Classification, J. Ilm. KURSOR, vol. 12, no. 2, pp. 8392, 2023. DOI: https://doi.org/10.21107/kursor.v12i2.349

[6] B. Sistaninejhad, H. Rasi, and P. Nayeri, A Review Paper about Deep Learning for Medical Image Analysis, Comput. Math. Methods Med., vol. 2023, 2023, doi: 10.1155/2023/7091301. DOI: https://doi.org/10.1155/2023/7091301

[7] M. M and S. P, COVID-19 infection prediction from CT scan images of lungs using Iterative Convolution Neural Network model, Adv. Eng. Softw., vol. 173, no. June, 2022, doi: 10.1016/j.advengsoft.2022.103214. DOI: https://doi.org/10.1016/j.advengsoft.2022.103214

[8] S. Surono, M. Yahya Firza Afitian, A. Setyawan, D. K. E. Arofah, and A. Thobirin, Comparison of CNN Classification Model using Machine Learning with Bayesian Optimizer, HighTech Innov. J., vol. 4, no. 3, pp. 531542, 2023, doi: 10.28991/HIJ-2023-04-03-05. DOI: https://doi.org/10.28991/HIJ-2023-04-03-05

[9] H. Naseri and A. A. Safaei, Diagnosis and prognosis of melanoma from dermoscopy images using machine learning and deep learning: a systematic literature review, BMC Cancer, vol. 25, no. 1, 2025, doi: 10.1186/s12885-024-13423-y. DOI: https://doi.org/10.1186/s12885-024-13423-y

[10] S. G. Yogita K. Desai, Tushar Jadhav, Suvarna Nirmal, Ishita Dutt, Suchita A. Chavan, Avinash Thakur, View of Diagnosis of Medical Images Using Convolutional Neural Networks.pdf, 2024. DOI: https://doi.org/10.52783/jes.3220

[11] B. Ji, Optimizing Medical Imaging with AI: Advanced Cnn and Gan Techniques for Enhanced Diagnostic Efficiency, ITM Web Conf., vol. 78, p. 02005, 2025, doi: 10.1051/itmconf/20257802005. DOI: https://doi.org/10.1051/itmconf/20257802005

[12] I. D. Mienye, T. G. Swart, G. Obaido, M. Jordan, and P. Ilono, Deep Convolutional Neural Networks in Medical Image Analysis: A Review, Inf., vol. 16, no. 3, pp. 128, 2025, doi: 10.3390/info16030195. DOI: https://doi.org/10.3390/info16030195

[13] I. Araf, A. Idri, and I. Chairi, Cost sensitive learning for imbalanced medical data: a review, vol. 57, no. 4. Springer Netherlands, 2024. doi: 10.1007/s10462-023-10652-8. DOI: https://doi.org/10.1007/s10462-023-10652-8

[14] E. Goceri, Medical image data augmentation: techniques, comparisons and interpretations, vol. 56, no. 11. Springer Netherlands, 2023. doi: 10.1007/s10462-023-10453-z. DOI: https://doi.org/10.1007/s10462-023-10453-z

[15] A. Larasati, S. Surono, A. Thobirin, and D. A. Dewi, Performance Analysis of Resampling Techniques for Overcoming Data Imbalance in Multiclass Classification, JUITA J. Inform., vol. 13, no. 1, pp. 5766, 2025, doi: 10.30595/juita.v13i1.25270. DOI: https://doi.org/10.30595/juita.v13i1.25270

[16] H. E. Kim, A. C. Linan, N. Santhanam, M. Jannesari, M. E. Maros, and T. Ganslandt, Transfer learning for medical image classification: a literature review, BMC Med. Imaging, pp. 113, 2022, doi: 10.1186/s12880-022-00793-7. DOI: https://doi.org/10.21203/rs.3.rs-844222/v1

[17] M. Liu, Y. Pei, M. Wu, and J. Wang, Focal Cosine-Enhanced EfficientNetB0: A Novel Approach to Classifying Breast Histopathological Images, Inf., vol. 16, no. 6, 2025, doi: 10.3390/info16060444. DOI: https://doi.org/10.3390/info16060444

[18] E. Gayatri and S. L. Aarthy, Reduction of overfitting on the highly imbalanced ISIC-2019 skin dataset using deep learning frameworks, J. Xray. Sci. Technol., vol. 32, no. 1, pp. 5368, 2024, doi: 10.3233/XST-230204. DOI: https://doi.org/10.3233/XST-230204

[19] D. N. T. Le, H. X. Le, L. T. Ngo, and H. T. Ngo, Transfer learning with class-weighted and focal loss function for automatic skin cancer classification, pp. 17.

[20] P. Tschandl, C. Rosendahl, and H. Kittler, Data Descriptor: The HAM 10000 dataset , a large collection of multi-source dermatoscopic images of common pigmented skin lesions, Nat. Publ. Gr., pp. 19, 2018, doi: 10.1038/sdata.2018.161. DOI: https://doi.org/10.1038/sdata.2018.161

[21] A. A. A. El-aziz, M. A. Mahmood, and S. A. El-ghany, A Robust EfficientNetV2-S Classifier for Predicting Acute Lymphoblastic Leukemia Based on Cross Validation, 2025. DOI: https://doi.org/10.3390/sym17010024

[22] M. Latha, P. S. Kumar, R. R. Chandrika, T. R. Mahesh, V. V. Kumar, and S. Guluwadi, Revolutionizing breast ultrasound diagnostics with EfficientNet B7 and Explainable, BMC Med. Imaging, 2024, doi: 10.1186/s12880-024-01404-3. DOI: https://doi.org/10.1186/s12880-024-01404-3

[23] A. Razaq, S. Ramzan, S. Jabbar, M. M. Iqbal, M. A. Habib, and U. Raza, A Deep Learning Framework for Early Parkinsons Disease Detection: Leveraging Spiral and Wave Handwriting Tasks with EfficientNetV2-S, Diagnostics, vol. 15, no. 21, pp. 125, 2025, doi: 10.3390/diagnostics15212795. DOI: https://doi.org/10.3390/diagnostics15212795

[24] M. Tan and Q. V Le, EfficientNetV2: Smaller Models and Faster Training, 2021.

[25] H. M. Hang Ma, Y.-H. Z. Hang Ma, B.-S. L. Yu-Hang Zhang, and W.-B. C. Bo-Si Liu, DETRs with Dynamic Contrastive Denoising Training for Smartphone Assembly Parts, , vol. 35, no. 3, pp. 175192, 2024, doi: 10.53106/199115992024063503013. DOI: https://doi.org/10.53106/199115992024063503013

[26] M. Motiejauskas and G. Dzemyda, CCC Publications EfficientNet Convolutional Neural Network with Gram Matrices Modules for Predicting Sadness Emotion, 2024. DOI: https://doi.org/10.15837/ijccc.2024.5.6697

[27] L. Deng, H. Suo, and D. Li, Deepfake Video Detection Based on EfficientNet-V2 Network, vol. 2022, 2022. DOI: https://doi.org/10.1155/2022/3441549

[28] S. Tummala, V. S. G. Thadikemalla, S. Kadry, M. Sharaf, and H. T. Rauf, EfficientNetV2 Based Ensemble Model for Quality Estimation of Diabetic Retinopathy Images from DeepDRiD, Diagnostics, vol. 13, no. 4, 2023, doi: 10.3390/diagnostics13040622. DOI: https://doi.org/10.3390/diagnostics13040622

[29] Z. Zhao, E. B. A. Bakar, N. B. A. Razak, and M. N. Akhtar, Corrosion image classification method based on EfficientNetV2, Heliyon, vol. 10, no. 17, p. e36754, 2024, doi: 10.1016/j.heliyon.2024.e36754. DOI: https://doi.org/10.1016/j.heliyon.2024.e36754

[30] . Kaiser, A. N. Gomez, and F. Chollet, Depthwise separable convolutions for neural machine translation, 6th Int. Conf. Learn. Represent. ICLR 2018 - Conf. Track Proc., 2018.

[31] M. Sandler, A. Howard, M. Zhu, and A. Zhmoginov, Sandler_MobileNetV2_Inverted_Residuals_CVPR_2018_paper.pdf, arXiv, pp. 45104520, 2018.

[32] C. Li, K. Liu, and S. Liu, A Survey of Loss Functions in Deep Learning, Mathematics, vol. 13, no. 15, p. 2417, 2025, doi: 10.3390/math13152417. DOI: https://doi.org/10.3390/math13152417

[33] J. T. D. C. J. Romero-, A comprehensive survey of loss functions and metrics in deep learning, 2025.

[34] M. Reyad and A. M. Sarhan, A modified Adam algorithm for deep neural network optimization, Neural Comput. Appl., vol. 35, no. 23, pp. 1709517112, 2023, doi: 10.1007/s00521-023-08568-z. DOI: https://doi.org/10.1007/s00521-023-08568-z

[35] R. V Kashyap, 111 A survey of deep learning optimizers - first and second order methods, 2023.

[36] Y. Wang, Z. Xiao, and G. Cao, A convolutional neural network method based on Adam optimizer with power-exponential learning rate for bearing fault diagnosis, pp. 666678, 2022, doi: 10.21595/jve.2022.22271. DOI: https://doi.org/10.21595/jve.2022.22271

[37] G. Zeng, Invariance Properties and Evaluation Metrics Derived from the Confusion Matrix in Multiclass Classification, no. 1997, 2025. DOI: https://doi.org/10.3390/math13162609

[38] C. A. Sugianto, D. Rohmayani, J. Fredricka, and M. Doheir, Interpretable Deep Learning Model for Grape Leaf Disease Classification Based on EfficientNet with Grad- CAM Visualization, pp. 143152, 2025.

[39] N. W. Utami, M. Adi, and P. Putra, A CNN-Based Information System for Balinese Dance Classification with Hyperparameter Optimization, vol. 6, no. 2, pp. 376390, 2025, doi: 10.59395/ijadis.v6i2.1402. DOI: https://doi.org/10.59395/ijadis.v6i2.1402