Skip to content
SDGs Center UDINUS

Evidence record · UDINUS · SDGs

Performance Evaluation of Oversampling Methods on Deep Learning-Based Skin Cancer Classification

Rektorat UDINUS 2023 Published 2026-10-30

Evidence record · UDINUS · SDGs

Unit
Rektorat UDINUS
Period
2023
Goals
Record
bd9fad00-af0b-4d70-943e-bb4800c21603
Published
2026-10-30

Universitas Dian Nuswantoro (UDINUS), through researchers Supriyanto, Catur; Salam, Abu; Zeniarja, Junta; and Wijaya, Adi, conducted an activity on the performance evaluation of oversampling methods for deep learning-based skin cancer classification. The activity addressed challenges in the early detection of skin cancer by comparing original data, augmented data, and SMOTE oversampled data and examining their effects on classification performance.

The activity used images of skin lesions from the MNIST Skin Cancer dataset, identified as HAM 10000. The dataset included both cancerous and benign cases. UDINUS researchers prepared original data, augmented data, and data processed with SMOTE oversampling. Data augmentation was applied to expand the dataset and increase the diversity of skin lesion features. SMOTE oversampling was applied to address class imbalance by generating synthetic samples for the under-represented group.

Using these dataset conditions, the researchers trained a Convolutional Neural Network model. Model performance was evaluated through accuracy, recall, precision, and F1-score. The comparison showed distinctions in performance among the original, augmented, and SMOTE oversampled datasets. The findings stated that data augmentation and SMOTE oversampling can significantly enhance the efficacy of skin cancer detection.

This UDINUS activity concerns health research on skin cancer detection and the evaluation of methods used in that research. It involved dataset preparation, data augmentation, SMOTE oversampling, and deep learning model evaluation for a skin cancer classification task.

Reference https://www.scopus.com/pages/publications/85178076637?origin=resultslist