<![CDATA[Skin health is an important aspect of maintaining quality of life, while skin cancer requires early detection to support appropriate treatment. This study explores the application of Explainable Artificial Intelligence (XAI), particularly LIME and Eigen-CAM, to interpret the results of binary skin cancer classification into benign and malignant classes. Classification was performed using a transfer learning approach with ResNet50, VGG16, EfficientNetB0, and MobileNetV2 architectures on a dataset of 2,000 skin images, with an equal number of samples in both classes. Model performance was evaluated using 5-fold cross-validation to obtain more consistent performance estimates across the test data. The experimental results showed that ResNet50 achieved the highest accuracy of 86.90%, followed by VGG16 at 85.90%, MobileNetV2 at 84.90%, and EfficientNetB0 at 84.85%. Visualization using LIME and Eigen-CAM highlighted image regions that contributed to the model's predictions, providing additional insights into the basis of its classification decisions. These findings indicate that XAI can serve as an interpretative approach for improving the understanding of model predictions in binary skin cancer classification.]]>
Copyrights © 2026
