Claim Missing Document
Check
Articles

Found 40 Documents
Search

Convolutional neural network hyperparameters for face emotion recognition using genetic algorithm Muhammad Sam'an; Safuan Safuan; Muhammad Munsarif
Indonesian Journal of Electrical Engineering and Computer Science Vol 33, No 1: January 2024
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/ijeecs.v33.i1.pp442-449

Abstract

The development of artificial intelligence in facial emotion recognition (FER) is rapidly growing and has been widely applied in various fields. Deep learning (DL) techniques with evolutionary algorithms have become the preferred choice for solving various security, health, gaming, and other related problems. This research proposes the use of a genetic algorithm (GA) as the main method to optimize hyperparameters in the convolutional neural network (CNN) model for FER. The required computation time is approximately 37 hours 57 minutes 55 seconds, with generation 3 taking the longest time at around 16 hours 45 minutes 4 seconds. However, generation 3 achieved an accuracy of 76.11%, which is the highest compared to other generations. The results indicate that the more generations are involved, the higher the achievable accuracy. Furthermore, the proposed CNN-GA model in this study outperforms previous models that have been examined. Thus, this study makes a significant contribution to improving the understanding of using GAs to optimize the performance of CNN models for FER.
Sistem Klasifikasi Tahu Non-Formalin Menggunakan Metode Random Forest Sefrina Ainun; Muhammad Munsarif; Muhammad Sam'an
JURNAL KOMPUTER DAN TEKNOLOGI INFORMASI Vol 1, No 2 (2023): Sistem Pengambilan Keputusan
Publisher : Universitas Muhammadiyah Semarang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26714/jkti.v1i2.12861

Abstract

Tahu formalin adalah salah satu jenis makanan yang mengandung bahan-bahan kimia yang dapat mengawetkan daripada tahu tanpa formalin. Penelitian ini bertujuan untuk mengklasifikasikan tahu formalin dan tahu tidak formalin. Penelitian ini menggunakan metode random forest yang merupakan bagian dari algoritma machine learning untuk klasifikasi, Penelitian ini mencoba menerapkan metode random forest pada dataset tahu formalin dengan jumlah dataset public. Setelah dilakukan beberapa tahapan dalam pengujian dengan metode random forest maka diperolah hasil akurasi 89%. Model random forest dikembangkan menjadi aplikasi web deteksi tahu non formalin dan tahu formalin yang berfungsi bagi masyarakat dalam meningkatkan pangan agar bebas konsumsi tahu non formalin.
Enhancing Agricultural Pest Detection with EfficientNetV2-L and Grad-CAM: A Comprehensive Approach to Sustainable Farming Agatra, Denaya Ferrari Noval; Cornella, Barisma Ami; Muza'in, Muhammad; Munsarif, Muhammad; Abdollahi, Jafar; Ilham, Ahmad
Journal of Intelligent Computing & Health Informatics Vol 5, No 1 (2024): March
Publisher : Universitas Muhammadiyah Semarang Press

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26714/jichi.v5i1.13959

Abstract

In modern agriculture, quickly identifying agricultural pests is essential for maintaining high crop yields and ensuring global food security. In diverse and dynamic agricultural environments, traditional pest detection methods exhibit reduced accuracy, limited scalability, and lack interpretability. In this study, EfficientNetV2-L and Grad-CAM were used to significantly enhance pest detection system performance and transparency. EfficientNetV2-L, a fast and resource-efficient model, excels particularly in computationally constrained environments. Traditional CNN models, including EfficientNetV2-L, are criticized as uninterpretable "black boxes" despite their high accuracy. To address this issue, Grad-CAM was used to generate salient maps that visually show the most influential areas of the input image in the model’s decision-making process. This combination not only provides superior pest detection accuracy but also provides actionable insights into the model’s predictions, which is an important feature for building trust among agricultural practitioners. Our experimental results show a 15% improvement in detection accuracy compared to conventional models, especially in identifying visually similar-looking pest species that are often misclassified. In addition, the enhanced interpretability provided by Grad-CAM has led to a deeper understanding of the model’s behaviour, enabling iterative adjustments and improvements that further enhance the reliability of the system. The practical implications of these findings are significant: this integrated model offers a robust solution that can be seamlessly applied to real-time agricultural monitoring systems. With the early detection and proper classification of pests, this model can be used as a more effective pest management strategy to minimize crop damage and increase agricultural productivity. This research not only advances the technological frontier of pest detection but also contributes to broader goals related to sustainable agriculture and food security. Future research will focus on expanding the applicability of this model across different agricultural contexts, improving its adaptability to different environmental conditions, and further optimizing its performance through advanced techniques such as transfer learning and ensemble methods.
Feature selection in P2P lending for default prediction using grey wolf optimization and machine learning Sam'an, Muhammad; Safuan, Safuan; Munsarif, Muhammad
Bulletin of Electrical Engineering and Informatics Vol 13, No 5: October 2024
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eei.v13i5.7651

Abstract

Online loan services like peer-to-peer (P2P) lending enable lenders to transact without bank intermediaries. Predicting which lenders are likely to default is crucial to avoid bankruptcy since lenders bear the risk of default. However, this task becomes challenging when the P2P lending dataset contains numer- ous features. The prediction performance could be improved if the dataset fea- tures are relevant. Hence, applying feature selection to remove redundant and irrelevant features is essential. This paper introduces a novel wrapper feature selection model to identify the optimal feature subset for predicting defaults in P2P lending. The proposed method includes two main phases: feature selection and classification. Initially, grey wolf optimization (GWO) is used to select the best features in P2P lending datasets. Then, the fitness function of GWO is as- sessed using ten different machine learning (ML) models. Experimental results indicate that the proposed model outperforms previous related work, achieving average accuracy, recall, precision, and F1-score of 96.77%, 80.73%, 97.52%, and 80.06%, respectively.
Convolution neural network hyperparameter optimization using modified particle swarm optimization Munsarif, Muhammad; Sam'an, Muhammad; Fahrezi, Andrian
Bulletin of Electrical Engineering and Informatics Vol 13, No 2: April 2024
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eei.v13i2.6112

Abstract

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 (LDWPSO), 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, CAE1, 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.
Peningkatan Kompetensi Guru Madrasah Ibtidayah Duren dan Sabilul Huda Bandungan melalui Pelatihan Pembelajaran Berbasis Teknologi Informasi Ilham, Ahmad; Fathurrohman, Akhmad; Sam'an, Muhammad; Safuan, Safuan; Munsarif, Muhammad; Assaffat, Luqman; Kindarto, Asdani; Ramadhani, Arfido; Adinullhaq, Juyus Muhammad; Febrianto, Febrianto; Nurmantoro, Irvan; Ardhani, Yevi Alviatul; Ariyanto, Nova
Jurnal Surya Masyarakat Vol 5, No 2 (2023): Mei 2023
Publisher : Universitas Muhammadiyah Semarang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26714/jsm.5.2.2023.264-269

Abstract

Madrasah Ibtidaiyah (MI) Duren Village and Sabilul Huda Jimbaran Bandungan District Semarang Regency want to produce quality graduates. However, the competence of teachers is still conventional learning aids so the learning process is not optimal. To answer this problem, the Department of Informatics, Faculty of Engineering at Universitas Muhammadiyah Semarang, Indonesia proposed information technology-based learning training activities for madrasah teachers. The purpose program is to strengthen human resources for teachers in MI Desa Duren and Sabilul Huda Jimbaran. The proposed program is divided into three learning schemes, (1) interactive presentation media, (2) online classroom learning, and (3) online learning evaluation. The results of this program are that the participants proved to be able to produce effective, elaborative, and interactive teaching materials based on information technology so that students are not bored and enthusiastic about following lessons in the classroom. It can be cancluded the program with the theme "Strengthening Teacher Competencies Through Information Technology-Based Learning Training" can overcome problems in Madrasah Ibtidaiyah (MI) Duren Village and Sabilul Huda Jimbaran Bandungan District Semarang Regency.
Penerapan algoritma convolutional neural network (cnn) untuk mengklasifikasi jenis kendaraan Wibowo, Bhanu Lintang; Caesarizky, Ovien Yoga; Hakim, Muhammad Ilham; Munsarif, Muhammad
JURNAL KOMPUTER DAN TEKNOLOGI INFORMASI Vol 2, No 2 (2024): Implementasi Sistem Cerdas
Publisher : Universitas Muhammadiyah Semarang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26714/jkti.v2i2.13938

Abstract

Dengan berkembangnya teknologi kendaraan otonom dan pengawasan lalu lintas cerdas, klasifikasi kendaraan menjadi elemen kunci dalam sistem transportasi modern. Artikel ini mengusulkan suatu pendekatan berbasis Convolutional Neural Network (CNN) untuk meningkatkan ketepatan dan efisiensi dalam klasifikasi jenis kendaraan dari data visual. Metode yang diusulkan menggunakan arsitektur CNN yang mendalam untuk mengekstraksi fitur-fitur penting dari citra kendaraan. Data pelatihan yang luas dan bervariasi digunakan untuk melatih model, yang kemudian diuji pada dataset independen untuk mengevaluasi performanya. Hasil eksperimen menunjukkan bahwa pendekatan ini dapat mencapai tingkat akurasi yang tinggi dalam klasifikasi kendaraan, bahkan dalam kondisi pencahayaan yang bervariasi dan latar belakang yang kompleks. Keuntungan utama dari metode ini adalah kemampuannya untuk secara otomatis mempelajari pola dan fitur- fitur diskriminatif dari citra kendaraan, membuatnya mampu mengatasi variasi dan kompleksitas dalam lingkungan praktis. Hasil dari penelitian ini memberikan kontribusi positif terhadap pengembangan sistem transportasi pintar dan meningkatkan keamanan serta efisiensi lalu lintas.
EfficientNet for Medical Image Classification: Performance vs. Efficiency in Skin Cancer Detection Purbandanu, Muhammad Wigig; Kurniawan, Arif; Yanuarta, Rizky; Munsarif, Muhammad; A. Awoseyi, Ayomikun
Journal of Intelligent Computing & Health Informatics Vol 5, No 2 (2024): September
Publisher : Universitas Muhammadiyah Semarang Press

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26714/jichi.v5i2.14338

Abstract

This study applies EfficientNetB2, a computationally efficient convolutional neural network (CNN), to improve the accuracy of skin cancer detection using the heterogeneous HAM10000 dataset. Skin cancer classification poses challenges, including overfitting and class imbalance, which we address through data augmentation, class weighting, and SMOTE (Synthetic Minority Over-sampling Technique). Our model achieved accuracy of 86%, precision of 0.87, recall of 0.85, and an AUC of 0.90. These results outperform comparable architectures, such as ResNet50 and GoogleNet, while maintaining lower computational complexity. The proposed model demonstrates high precision in detecting actinic keratoses and basal cell carcinoma, which require timely treatment, but faces difficulties in differentiating melanoma from benign nevi because of their similar visual appearance. This study highlights the potential of EfficientNetB2 for real-world deployment in resource-limited settings, such as mobile health applications and telemedicine platforms. Future research will focus on integrating attention mechanisms and exploring cross-dataset validation to enhance model generalizability and performance.
Pemberdayaan Guru melalui Pelatihan Pemanfaatan Kecerdasan Buatan (AI) untuk Meningkatkan Kualitas Pembelajaran di Era Digital Muhammad Munsarif; Muhammad Sam’an; Safuan Safuan
ASPIRASI : Publikasi Hasil Pengabdian dan Kegiatan Masyarakat Vol. 3 No. 1 (2025): ASPIRASI : Publikasi Hasil Pengabdian dan Kegiatan Masyarakat 
Publisher : Asosiasi Periset Bahasa Sastra Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61132/aspirasi.v3i1.1379

Abstract

The development of competent human resources (HR) is a top priority for improving the quality of life in society. In education, teachers' ability to enhance their skills is key to successful learning. However, many teachers struggle to adapt teaching methods to the demands of the digital era, especially in utilizing Artificial Intelligence (AI) technology. AI offers innovative solutions such as personalized learning and adaptive platforms that enhance the effectiveness of teaching and learning processes. This study examines the implementation of AI-based learning models and teacher training programs in Muhammadiyah schools in Semarang. Using the Training of Trainer (ToT) approach, the program equips teachers with skills such as using adaptive learning tools and AI-based evaluation systems. The study employs a mixed-methods approach involving practical training and result evaluation. The findings reveal improved digital literacy and confidence among teachers in using AI for interactive learning. In conclusion, AI has great potential to transform education, but its success depends on comprehensive, ethical, and sustainable strategies. The ToT program serves as a strategic step to empower teachers to face educational challenges in the digital era.
Applicaton Of Mathematics In Big Data Analysis To Support Strategic Decision Munsarif, Muhammad; Walid, Abul; Sari, Nila Kartika
Aksioma Education Journal Vol. 1 No. 4 (2024): December-AEJ
Publisher : PT. Anagata Sembagi Education

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.62872/tq97je41

Abstract

This study aims to investigate the application of mathematical models in big data Analysis and their impact on strategic decision making in various industrial sectors. Using a quantitative approach to the survey, data was collected from 190 respondents from the technology, finance, manufacturing and healthcare sectors. The results showed that the application of mathematical models, such as predictive algorithms and machine learning, contributed significantly to improving the quality of strategic decisions. The study also identified that variables such as human resource competence and technological infrastructure moderate the relationship between big data Analysis and effective decision-making. The technology and finance sectors have proven to benefit the most from the application of math-based big data Analytics, with benefits seen in improved market prediction, risk management, and operational optimization. The findings underscore the importance of integrating mathematical models in data analysis to support data-driven decision-making in the digital age.
Co-Authors A. Awoseyi, Ayomikun Abdollahi, Jafar Abul Walid Adi Nugroho Adinullhaq, Juyus Muhammad Afan Arga Ahyana Agatra, Denaya Ferrari Noval Ahmad Ilham Ahmad Ilham Ahmad Ilham Ahmad Miftakhul Akhmad Fathurohman Akhmad Fathurrohman Akhmad Fathurrohman Alfi Widianto Alif Hardika Andrian Fahrezi Ardhani, Yevi Alviatul Arfido Ramadhani Arianty, Alya Dwi ARIF KURNIAWAN Ariyanto, Nova Arnold Adimabua Ojugo Asdani Kindarto Asdani Kindarto Asiva Khoirunisa Budi Santosa Budi Santosa Caesarizky, Ovien Yoga Cornella, Barisma Ami De Rosal Ignatius Moses Setiadi Devy Maharja Pristya Dhendra Marutho Dhendra Marutho Edi Noersasongko Edy Winarno Eny Winaryati Eny Winaryati Ernawati Ernawati Fahrezi, Andrian Faiq Madani Fajar Bima Laksono Febrianto Febrianto Febrianto Febrianto, Febrianto Galih Ratna Hakim, Muhammad Ilham Hannan Isnaen, Muhammad Ilham, Ahmad Irvan Nurmantoro Irvan Nurmantoro Iwan Junaedi Janu Yogi Kurnia Javier Rayhan Jian Janata Juyus Muhammad Adinullhaq Kafitra Marna Ibrahim Kholifatul Afwah Kindarto, Asdani Ku Ruhana Ku-Mahamud Ku-Mahamud, Ku Ruhana Lukman Assaffat Lukman Assaffat Luqman Assaffat Luqman Assaffat M. Amin Syaifani Mahmudah Mahmudah Mardiana Mardiana Meike Amelia Hanifah Melda Saputri Mila Moch Arief Soeleman Mochammad Arief Soeleman Muhammad Sam'an Muhammad Sam'an Muhammad Sam’an Muhammad Sam’an Muhammad Sam’an Muhammad Wahyu Anggana Muthia Aulia Rahma Muza'in, Muhammad Nila Kartika Sari Norshuhani Zamin Norshuhani Zamin Nova Ariyanto Nur Faizah Nurmantoro, Irvan Nurul Farah Izzatuna Okta Wijaya, Risky Pulung Nurtantio Andono Purbandanu, Muhammad Wigig Ramadhani, Arfido Ramadhani, Rima Dias Richmond Ampah-Mensah Rima Dias Ramadhani Safuan Safuan Safuan Safuan Safuan Safuan Safuan Safuan Safuan Safuan Salsabila Rahma Saman, Muhammad Samsudi Raharjo Samsudi Raharjo Sefrina Ainun Shabrina Mirza Siti Muzdalifah Suwahono Suwahono Teguh Firmasyah Tirta Ramadhani Utomo Wendy Sarasjati Wendy Sarasjati Wibowo, Bhanu Lintang Yanuarta, Rizky Yesika Maya Ocktarani Yevi Alviatul Ardhani