Year: 2025 | Month: December | Volume: 12 | Issue: 12 | Pages: 684-691
DOI: https://doi.org/10.52403/ijrr.20251270
Optimization of Convolutional Neural Network-Based Classification Using EfficientNet-B1
Muhammad Fachri Mahyudin1, Maman Somantri1, Oky Dwi Nurhayati2
1Department of Electrical Engineering, 2Department of Computer Engineering,
Universitas Diponegoro, Semarang, Indonesia.
Corresponding Author: Muhammad Fachri Mahyudin
ABSTRACT
Brain cancer is a life-threatening disease with a global mortality count reaching 241,037 cases, with Asia recording the highest number of deaths. Advances in artificial intelligence (AI) and machine learning offer significant opportunities to improve the accuracy and consistency of diagnosis through MRI image analysis. Convolutional Neural Networks (CNNs) have been widely used in cancer detection due to their ability to automatically extract features and perform high-accuracy image classification. This thesis employs the EfficientNet B1 model because its compound scaling architecture optimally balances network depth, width, and resolution. This design enables the model to achieve high computational efficiency, operate smoothly on various hardware systems, and still maintain strong accuracy performance. These characteristics make EfficientNet B1 particularly suitable for identifying complex patterns in brain MRI images. This research focuses on optimizing and evaluating EfficientNet B1 for brain cancer detection tasks, emphasizing both accuracy and computational efficiency. The experimental results show that the model achieved an accuracy of 0.9734, confirming its effectiveness in brain cancer classification. These findings highlight the potential of EfficientNet B1 as a fast, accurate, and practical model for AI-based diagnostic support systems.
Keywords: Brain Cancer, MRI, CNN, EFFICIENTNET-B1
[PDF Full Text]