Optimalisasi Hyper Parameter Convolutional Neural Networks Menggunakan Ant Colony Optimization

Penulis

  • Fian Yulio Santoso Universitas Kristen Satya Wacana, Salatiga
  • Eko Sediyono Universitas Kristen Satya Wacana, Salatiga
  • Hindriyanto Dwi Purnomo Universitas Kristen Satya Wacana, Salatiga

DOI:

https://doi.org/10.25126/jtiik.20241127105

Kata Kunci:

auto-tuning hyperparameter, deep neural network, ant colony, genetic algorithm

Abstrak

Berbagai bidang, termasuk pertanian dan kesehatan, mengalami masalah klasifikasi citra yang dapat diatasi melalui beberapa metode. Salah satu metode tersebut menggabungkan convolutional neural networks (CNN) dengan deep learning, tetapi hyperparameter, seperti fungsi loss, fungsi aktivasi, dan optimizers, memengaruhi kinerjanya. Hyperparameter ini memerlukan pengoptimalan, dan metode yang ada, seperti algoritma genetika dan pengoptimalan ant colony, dapat digunakan untuk tujuan ini. Pengoptimalan ant colony terbukti efektif dalam mengoptimalkan deep learning, dan penelitian ini berkontribusi pada penyetelan otomatis berbagai hyperparameter menggunakan ant colony untuk klasifikasi gambar. Pada penelitian ini menggunakan dataset MNIST yang bertujuan untuk mengidentifikasi digit pada citra. Dataset yang digunakan terbagi menjadi 2, dataset pelatihan dan dataset validasi. Dataset pelatihan terdiri dari 33.600 gambar, dan dataset validasi terdiri dari 8.400 gambar. Hasil menunjukkan bahwa optimasi ant colony mencapai akurasi 97,46% dengan data validasi dan 99,69% dengan data pelatihan, yang mengungguli algoritma genetika dengan akurasi masing-masing 94,60% dan 97,59% dengan data validasi dan pelatihan. Selain itu, pengoptimalan ant colony membutuhkan waktu 27,94 detik untuk dilatih, sedangkan algoritme genetika membutuhkan 22,25 detik.

 

Abstract

Various fields, including agriculture and health, have encountered image classification problems that can be addressed through several methods. One such method combines convolutional neural networks (CNN) with deep learning, but hyperparameters, such as loss functions, activation functions, and optimizers, influence its performance. These hyperparameters require optimization, and existing methods, such as genetic algorithms and ant colony optimization, can be utilized for this purpose. Ant colony optimization has shown to be effective in optimizing deep learning, and this research contributes to automatic tuning of various hyperparameters using ant colonies for image classification. In this study using the MNIST dataset, which aims to identify the digits in the image. The dataset used is divided into 2, training dataset and validation dataset. The training dataset consists of 33,600 images, and the validation dataset consists of 8,400 images. The results indicate that ant colony optimization achieves an accuracy of 97.46% with validation data and 99.69% with training data, which outperforms genetic algorithms with an accuracy of 94.60% and 97.59% with validation and training data, respectively. Additionally, ant colony optimization takes 27.94 seconds to train, while the genetic algorithm requires 22.25 seconds.

Downloads

Download data is not yet available.

Referensi

AKBARI, A., AWAIS, M., BASHAR, M. AND KITTLER, J., 2021. How Does Loss Function Affect Generalization Performance of Deep Learning? Application to Human Age Estimation. In: International Conference on Machine Learning. [online] International Conference on Machine Learning. PMLR. pp.141–151. Available at: <https://proceedings.mlr.press/v139/akbari21a.html> [Accessed 14 March 2022].

AROWOLO, M.O., AWOTUNDE, J.B., AYEGBA, P. AND SULYMAN, S.O.H., 2022. Relevant gene selection using ANOVA-ant colony optimisation approach for malaria vector data classification. International Journal of Modelling, Identification and Control, [online] 41(1/2), p.12. https://doi.org/10.1504/ijmic.2022.127093.

ATTIQUE KHAN, M., MAJID, A., HUSSAIN, N., ALHAISONI, M., ZHANG, Y.-D., KADRY, S. AND NAM, Y., 2021. Multiclass stomach diseases classification using deep learning features optimization. Computers, materials & continua, [online] 67(3), pp.3381–3399. https://doi.org/10.32604/cmc.2021.014983.

CAI, L., GAO, J. AND ZHAO, D., 2020. A review of the application of deep learning in medical image classification and segmentation. Annals of translational medicine, [online] 8(11), p.713. https://doi.org/10.21037/atm.2020.02.44.

COLIANNI, S., 2017. MNIST as .jpg. Available at: <https://www.kaggle.com/scolianni/mnistasjpg> [Accessed 28 October 2022].

DORIGO, M. AND STÜTZLE, T., 2019. Ant colony optimization: Overview and recent advances. In: Handbook of Metaheuristics, International series in operations research & management science. [online] Cham: Springer International Publishing. pp.311–351. https://doi.org/10.1007/978-3-319-91086-4_10.

FUJIYOSHI, H., HIRAKAWA, T. AND YAMASHITA, T., 2019. Deep learning-based image recognition for autonomous driving. IATSS research, [online] 43(4), pp.244–252. https://doi.org/10.1016/j.iatssr.2019.11.008.

HE, T., LIU, Y., YU, Y., ZHAO, Q. AND HU, Z., 2020. Application of deep convolutional neural network on feature extraction and detection of wood defects. Measurement, [online] 152(107357), p.107357. https://doi.org/10.1016/j.measurement.2019.107357.

KHAN, A., ATTIQUE KHAN, M., YOUNUS JAVED, M., ALHAISONI, M., TARIQ, U., KADRY, S., CHOI, J.-I. AND NAM, Y., 2022. Human gait recognition using deep learning and improved ant colony optimization. Computers, materials & continua, [online] 70(2), pp.2113–2130. https://doi.org/10.32604/cmc.2022.018270.

KUMAR, A., SARKAR, S. AND PRADHAN, C., 2020. Malaria disease detection using CNN technique with SGD, RMSprop and ADAM optimizers. In: Studies in Big Data. [online] Cham: Springer International Publishing. pp.211–230. https://doi.org/10.1007/978-3-030-33966-1_11.

MURUGAN, A., NAIR, S.A.H. AND KUMAR, K.P.S., 2019. Detection of Skin Cancer Using SVM, Random Forest and kNN Classifiers. Journal of medical systems, [online] 43(8), p.269. https://doi.org/10.1007/s10916-019-1400-8.

MYTHILI, K. AND RANGARAJ, R., 2021. Crop Recommendation for Better Crop Yield for Precision Agriculture Using Ant Colony Optimization with Deep Learning Method. Annals of the Romanian Society for Cell Biology, [online] pp.4783–4794. Available at: <https://www.annalsofrscb.ro/index.php/journal/article/view/3024> [Accessed 15 February 2023].

NEWTON, D., PASUPATHY, R. AND YOUSEFIAN, F., 2018a. Recent trends in stochastic gradient descent for machine learning and big data. In: 2018 Winter Simulation Conference (WSC). [online] 2018 Winter Simulation Conference (WSC). IEEE. https://doi.org/10.1109/wsc.2018.8632351.

NEWTON, D., PASUPATHY, R. AND YOUSEFIAN, F., 2018b. Recent trends in stochastic gradient descent for machine learning and big data. In: 2018 Winter Simulation Conference (WSC). [online] 2018 Winter Simulation Conference (WSC). IEEE. https://doi.org/10.1109/wsc.2018.8632351.

VANI, S. AND RAO, T.V.M., 2019. An experimental approach towards the performance assessment of various optimizers on convolutional neural network. In: 2019 3rd International Conference on Trends in Electronics and Informatics (ICOEI). [online] 2019 3rd International Conference on Trends in Electronics and Informatics (ICOEI). IEEE. https://doi.org/10.1109/icoei.2019.8862686.

WANG, P., FAN, E.N. AND WANG, P., 2021. Comparative analysis of image classification algorithms based on traditional machine learning and deep learning. Pattern recognition letters, [online] 141, pp.61–67. https://doi.org/10.1016/j.patrec.2020.07.042.

ZAHEER, R. AND SHAZIYA, H., 2019. A study of the optimization algorithms in deep learning. In: 2019 Third International Conference on Inventive Systems and Control (ICISC). [online] 2019 Third International Conference on Inventive Systems and Control (ICISC). IEEE. https://doi.org/10.1109/icisc44355.2019.9036442.

Unduhan

Diterbitkan

26-08-2024

Terbitan

Bagian

Ilmu Komputer

Cara Mengutip

Optimalisasi Hyper Parameter Convolutional Neural Networks Menggunakan Ant Colony Optimization. (2024). Jurnal Teknologi Informasi Dan Ilmu Komputer, 11(2), 243-248. https://doi.org/10.25126/jtiik.20241127105