Bulletin of Electrical Engineering and Informatics Vol. 13, No. 2, April 2024, pp. 1268~1275 ISSN: 2302-9285, DOI: 10.11591/eei.v13i2.6112  1268 Journal homepage: http://beei.org Convolution neural network hyperparameter optimization using modified particle swarm optimization Muhammad Munsarif, Muhammad Sam’an, Andrian Fahrezi Department of Informatics, Universitas Muhammadiyah Semarang, Semarang, Indonesia Article Info ABSTRACT Article history: Received Feb 28, 2023 Revised Jul 16, 2023 Accepted Sep 11, 2023 Based on the literature review, a convolutional neural network (CNN) is one of the deep learning techniques most often used for classification problems, especially image classification. Various approaches have been proposed to improve accuracy performance. In CNN architecture, parameter determination is very influential on accuracy performance. Particle swarm optimization (PSO) is a type of metaheuristic algorithm widely used for hyperparameter optimization. PSO convergence is faster than genetic algorithm (GA) and attracts many researchers for further studies such as genetic algorithms and ant colony. In PSO, determining the value of the weight parameter is very influential on accuracy. Therefore, this paper proposes CNN hyperparameter optimization using modified PSO with linearly decreasing randomized weight. The experiments use the modified National Institute of Standards and Technology (MNIST) dataset. The accuracy of the proposed method is superior, and the execution time is slower to random search. In epoch 1, epoch 3, and epoch 5, the proposed method is superior to baseline CNN, linearly decreasing weight PSO (LDW- PSO), and RL-based optimization algorithm (ROA). Meanwhile, the accuracy performance of the proposed method is superior to previous studies, namely LeNet-1, LeNet-2, LeNet-3, PCANet-2, RANDNet-2, CAE- 1, CAE-2, and bee colony. Otherwise, lost to PSO-CNN, distributed PSO (DPSO), recurrent CNN, and CNN-PSO. However, the four methods have a complex architecture and wasteful execution time. Keywords: Convolutional neural network Handwritten digit Hyperparameter optimization Modified National Institute of Standards and Technology Particle swarm optimization This is an open access article under the CC BY-SA license. Corresponding Author: Muhammad Munsarif Department of Informatics, Universitas Muhammadiyah Semarang Jl. Kedung Mundu Raya No. 18, Semarang, 50273 Jawa Tengah, Indonesia Email: m.munsarif@unimus.ac.id 1. INTRODUCTION Literally, convolutional neural network (CNN) is recognized as one of the most powerful deep learning models for accurately predicting handwritten recognition with high accuracy [1], [2]. Compared to other deep learning models such as deep neural network (DNN), recurrent neural network (RNN), and artificial neural network (ANN) in image classification [3]–[5], CNN exhibits higher accuracy and faster execution time. The CNN architecture consists of two main components Figure 1: the convolution layer used for feature extraction and the Fully connected layer used for the classification process. The feature extraction process plays a crucial role in determining prediction accuracy, leading many researchers to explore various CNN architectural models that can achieve optimal performance [6]. One particular focus of research on CNN architectural models is the optimization of parameters using a hyperparameter approach [7].