Evaluasi Efektivitas Teknik Regularisasi Dalam Mengurangi Overfitting Pada Model CNN
Abstract
Penelitian ini bertujuan mengevaluasi dan membandingkan efektivitas berbagai teknik regularisasi seperti regularisasi L1 dan L2, dropout, dan augmentasi data, baik secara terpisah maupun kombinasi, dalam mengatasi overfitting pada model Convolutional Neural Network (CNN) dalam skenario dataset terbatas. Keterbatasan dataset merupakan tantangan utama yang menyebabkan model CNN cenderung mengalami overfitting, di mana performa pada data pelatihan 97.95% akurasi jauh melebihi akurasi validasi 67%. Penelitian ini menggunakan arsitektur CNN dasar yang konsisten dan dataset CIFAR-10. Hasil pengujian teknik regularisasi tunggal menunjukkan bahwa augmentasi data adalah teknik yang paling optimal pada pengujian terpisah. Model dengan augmentasi data mencapai akurasi validasi tertinggi 78.18% dan kesenjangan generalisasi terendah 2.31% di antara semua teknik yang diuji. Sementara itu, ditemukan bahwa penggunaan tingkat regularisasi yang terlalu ekstrem pada teknik regularisasi L1/L2 dapat menyebabkan underfitting karena bobot dipaksa mendekati nol sehingga model kehilangan kapasitas belajar. Pencapaian kinerja model yang paling superior diperoleh melalui pendekatan kombinasi. Kombinasi antara augmentasi data dan regularisasi L2 menghasilkan akurasi validasi tertinggi sebesar 79.89% dengan kesenjangan generalisasi paling kecil, yaitu 0.38%. Dengan demikian, disimpulkan bahwa pendekatan kombinasi teknik regularisasi adalah strategi paling efektif untuk meningkatkan generalisasi model CNN pada lingkungan dengan dataset terbatas.
Keywords
Full Text:
PDFReferences
D. H. Hubel and T. N. Wiesel, “Receptive fields, binocular interaction and functional architecture in the cat’s visual cortex,” J Physiol, vol. 160, no. 1, pp. 106–154, 1962, doi: 10.1113/jphysiol.1962.sp006837.
Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, “Gradient-based learning applied to document recognition,” Proceedings of the IEEE, vol. 86, no. 11, pp. 2278–2323, 1998, doi: 10.1109/5.726791.
L. Chen, S. Li, Q. Bai, J. Yang, S. Jiang, and Y. Miao, “Review of Image Classification Algorithms Based on Convolutional Neural Networks,” pp. 1–51, 2021.
A. Younesi, M. Ansari, M. Fazli, A. Ejlali, M. Shafique, and J. Henkel, “A Comprehensive Survey of Convolutions in Deep Learning: Applications, Challenges, and Future Trends,” IEEE Access, vol. 12, pp. 41180–41218, 2024, doi: 10.1109/ACCESS.2024.3376441.
Z. Xia, “Overfitting of CNN model in cifar-10: Problem and solutions,” Applied and Computational Engineering, vol. 37, no. 1, pp. 212–221, Feb. 2024, doi: 10.54254/2755-2721/37/20230511.
A. Yan and N. Wiliani, “Analisis Pengaruh Dropout Layer pada Convolutional Neural Networks untuk Klasifikasi Gambar Sticky notes,” vol. 13, no. 3, pp. 707–714, 2025.
L. Brigato and L. Iocchi, “A Close Look at Deep Learning with Small Data,” Mar. 2020, [Online]. Available: http://arxiv.org/abs/2003.12843
M. Faris and A. Hakim, “CNN model for the small data in vehicle miniature classification,” 2023 1st International Conference on Advanced Engineering and Technologies (ICONNIC), no. October, pp. 368–372, 2023, doi: 10.1109/ICONNIC59854.2023.10467559.
C. Aliferis and G. Simon, Overfitting, Underfitting and General Model Overconfidence and Under-Performance Pitfalls and Best Practices in Machine Learning and AI. 2024. doi: 10.1007/978-3-031-39355-6_10.
M. Kitahashi and H. Handa, “Application of Evolutionary Multiobjective Optimization in L1-Regularization of CNN,” 2019 IEEE Congress on Evolutionary Computation, CEC 2019 - Proceedings, pp. 2129–2135, 2019, doi: 10.1109/CEC.2019.8789899.
O. Harris, M. Andrews, P. G. Allen, and C. Science, “Effects and results of dropout layer in reducing overfitting with convolutional Neural Networks ( CNN ),” vol. 13, no. 02, pp. 836–853, 2024, [Online]. Available: https://doi.org/10.30574/wjaets.2024.13.2.0584
T. Kumar, R. O. B. Brennan, S. Member, A. Mileo, and M. Bendechache, “Image Data Augmentation Approaches : A Comprehensive Survey and Future Directions,” vol. 12, no. September, 2024, [Online]. Available: https://doi.org/10.48550/arXiv.2301.02830
A. Mumuni and F. Mumuni, “Data augmentation: A comprehensive survey of modern approaches,” Array, vol. 16, no. November, p. 100258, 2022, doi: 10.1016/j.array.2022.100258.
O. M. Khanday, S. Dadvandipour, and M. A. Lone, “Effect of filter sizes on image classification in CNN: A case study on CFIR10 and fashion-MNIST datasets,” IAES International Journal of Artificial Intelligence, vol. 10, no. 4, pp. 872–878, 2021, doi: 10.11591/ijai.v10.i4.pp872-878.
Purwono, A. Ma’arif, W. Rahmaniar, H. I. K. Fathurrahman, A. Z. K. Frisky, and Q. M. U. Haq, “Understanding of Convolutional Neural Network (CNN): A Review,” International Journal of Robotics and Control Systems, vol. 2, no. 4, pp. 739–748, 2022, doi: 10.31763/ijrcs.v2i4.888.
I. Nathan, “CNN Complexity Reduction,” pp. 1–25, 2022, [Online]. Available: https://arxiv.org/html/2501.15547v1
T. Measures, “Evaluating the Learning Procedure of CNNs through a Sequence of Prognostic Tests Utilising Information Theoretical Measures,” 2022, [Online]. Available: https://doi.org/10.3390/e24010067
B. A. Skourt, A. El Hassani, and A. Majda, “Mixed-pooling-dropout for convolutional neural network regularization,” Journal of King Saud University - Computer and Information Sciences, vol. 34, no. 8, pp. 4756–4762, 2022, doi: 10.1016/j.jksuci.2021.05.001.
Z. Zhu, “Systematic Optimization of Overfitting Problem in Machine Learning,” Highlights in Science, Engineering and Technology, vol. 111, pp. 353–359, 2024, doi: 10.54097/3tkzrj84.
L. Nanni, M. Paci, and S. Brahnam, “Comparison of Different Image Data Augmentation Approaches,” 2021, [Online]. Available: https://doi.org/10.3390/jimaging7120254
N. Nakamura and D. Maeda, “Efficient Parameters for Rotation Processing of Data Augmentation,” no. c, pp. 62–65, 2020, [Online]. Available: https://www.iaria.org/conferences2020/fileseKNOW20/eKNOW_60042.pdf
DOI: http://dx.doi.org/10.36448/expert.v15i2.4676
Refbacks
- There are currently no refbacks.
EXPERT: Jurnal Manajemen Sistem Informasi dan Teknologi
Published by Pusat Studi Teknologi Informasi, Fakultas Ilmu Komputer, Universitas Bandar Lampung
Gedung M Lt.2 Pascasarjana Universitas Bandar Lampung
Jln Zainal Abidin Pagaralam No.89 Gedong Meneng, Rajabasa, Bandar Lampung,
LAMPUNG, INDONESIA
Indexed by:
This work is licensed under a Creative Commons Attribution 4.0 International License.






