Evidence record · UDINUS · SDGs
- Unit
- Rektorat UDINUS
- Period
- 2024
- Goals
- Record
- bf2a57dd-6931-4787-a6db-6d60184022a7
- Published
- 2026-10-30
Universitas Dian Nuswantoro (UDINUS), through the participation of Nugroho, Wahyu Adi; Supriyanto, Catur; Shidik, Guruh Fajar; and Pujiono, conducted a study on modified rectified linear unit (ReLU) activation functions in deep-learning models and explainable artificial intelligence (XAI) techniques for breast cancer subtype classification. The activity addressed breast cancer as a serious condition that presents a considerable risk to life when it is not identified in its initial stages. It also considered the limitations of conventional medical examination in accuracy and reliability.
The study aimed to improve the performance of a convolutional neural network (CNN) model by modifying the ReLU activation function to distinguish eight breast cancer subtypes. The activity examined deep-learning methodologies for automated breast cancer diagnosis, with particular attention to CNN models. Based on the experiment results, the DenseNet201 model using the Less-NegativeReLU modified activation function achieved the best reported performance. Its reported accuracy was 96.04% with α=0.091, precision was 96.45% with α=0.03, recall was 96.04% with α=0.09, and F1-score was 96.14% with α=0.03.
UDINUS also employed XAI techniques to improve understanding of model prediction results and to enhance clinician confidence in implementation. The findings demonstrate that the proposed method effectively enhanced and optimized CNN model performance. By addressing both classification performance and interpretability, the activity relates to the stated goal concerning health and supports the institutional contribution of UDINUS to research on breast cancer diagnosis.