Claim Missing Document
Check
Articles

Found 8 Documents
Search

Deteksi Tumor Otak Dengan CNN Resnet-152 Aji Digdoyo; Tri Surawan; Adhitio Satyo Bayangkari Karno; Dyah Ruri Irawati; Yasin Effendi
Jurnal Teknologi Vol 9, No 2 (2022): Jurnal Teknologi
Publisher : Universitas Jayabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31479/jtek.v9i2.128

Abstract

Penyakit tumor di Indonesia menduduki tingkat kematian terbesar ke-5 setelah diabetes, stroke, ginjal dan darah tinggi. Kurangnya penanganan dini, jumlah paramedis dan peralatan yang minim adalah penyebab utama tingginya tingkat kematian. Artificial Intelligence (AI) memiliki potensi tinggi untuk berkontribusi membantu pasien dan para medis mendiagnosa tumor secara langsung, cepat dan murah. Salah satu metode AI dipergunakan dalam penelitian ini adalah CNN dengan arsitektur ResNet-152. Dengan melakukan training dan validasi sejumlah 2.870 image menghasilkan nilai akurasi masing-masing 99% dan 81%. Untuk lebih memastikan hasil yang diinginkan, maka model yang diperoleh dari training dipergunakan kembali untuk pengujian, dengan hasil nilai akurasi sebesar 96% dan akurasi untuk tiap kelas adalah glioma (97%), meningioma (95%), no_tumor (98%) dan pituaty (96%).
Mengatasi Ketimpangan Data Deep Neural Network dengan Pelipatan Fitur Data Klasifikasi Spektroskopi Darah Widi Hastomo; Adhitio Satyo Bayangkari Karno; Sutarno Sutarno; Dodi Arif; Eka Sally Moreta; Sudjiran Sudjiran
Sang Pencerah: Jurnal Ilmiah Universitas Muhammadiyah Buton Vol 8 No 2 (2022): Sang Pencerah: Jurnal Ilmiah Universitas Muhammadiyah Buton
Publisher : Lembaga Penelitian dan Pengabdian Masyarakat Universitas Muhammadiyah Buton

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (1673.005 KB) | DOI: 10.35326/pencerah.v8i2.2251

Abstract

Permasalahan utama dalam penelitian ini adalah ketimpangan data masukan menghasilkan dampak negatif yang signifikan terhadap hasil prediksi dari model Deep Neural Network (DNN). Kemampuan klasifikasi DNN sangat akurat hanya untuk dataset yang berimbang, namun DNN pada awalnya tidak di rancang untuk menangani ketimpangan data. Ketimpangan data merupakan hal yang sering dijumpai dalam dunia nyata, menjadikan ini sebagai tantangan besar dalam prediksi klasifikasi menggunakan model DNN. Penelitian ini berfokus untuk memprediksi tingkat kandungan kolesterol tinggi, kolesterol rendah dan hemoglobin, menggunakan data kasus di kompetisi Zindi Blood Spectroscopy Classification Challenge. Dengan melakukan analisa data, cleansing outlier, fine tunning, model neural network, jaringan pengelompokan data target dengan kategori sejenis, urutan pemrosesan, pemilihan nilai pelipatan (7 pelipatan) yang tepat terhadap data input train dan data test serta epoch 60, dapat meningkatkan hasil nilai score prediksi yang cukup tinggi sebesar 0.94594.
Exloratory Data Analysis Untuk Data Belanja Pelanggan dan Pendapatan Bisnis Widi Hastomo; Adhitio Satyo Bayangkari Karno; Sudjiran; Dodi Arif; Eka Sally Moreta
Infotekmesin Vol 13 No 2 (2022): Infotekmesin: Juli, 2022
Publisher : P3M Politeknik Negeri Cilacap

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35970/infotekmesin.v13i2.1547

Abstract

A more quantifiable perspective is assuming the role of mechanistic management in an effort to enhance business based on its capacity to transform data into knowledge and insight. The industry has not completely supported its business strategy also with driven data. Using a transaction dataset taken from one of the Kaggle.com challenges, this experiment attempts to determine consumer spending patterns and Retail Fashion business revenues (H&M Personalized Fashion Recommendations). The results of the experiment are the number of transactions based on customer age, the most sales product and one-time purchased item, and the type of product that generates the highest and smallest income. The approach employed is EDA using the Python language. In order for businesses to generate analytical findings that provide future perspectives and to help identify the gap by delivering analytical results in the form of suggestions that can be perpetuated, the findings of this experiment are intended to support the capabilities of simulation. The challenge in this experiment is the abundance of datasets, which necessitates a suitable operating environment.
MobilenetV2 Architecture To Detect Covid-19 X-Ray Imagery Widi Hastomo; Adhitio Satyo Bayangkari Karno; Ellya Sestri; Eva Karla; Stevianus Stevianus; Dodi Arif
Justek : Jurnal Sains dan Teknologi Vol 5, No 2 (2022): November
Publisher : Unversitas Muhammadiyah Mataram

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31764/justek.v5i2.11820

Abstract

Abstract:  The COVID-19 pandemic has hit all over the world, in the last two years and has changed the pace, structure and nature of social life. This study aims to detect COVID-19 using a chest x-ray image dataset sourced from kaggle.com, which is divided into 4 categories. The proposed method is CNN with MobileNetV2 architecture, by dividing 80% train data and 20% test data into 224x224 and batch size 32. The optimizer uses SGD, lr 0.005, momentum 0.9 and epoch 20. The results of the study with the achievement of precision values for the covid category 0.99, lung opacity 0.98, normal 0.96 and viral pneumonia category reached 0.99. Further studies can use the development of the CNN model and can try with other optimizers.Abstrak: Pandemi covid-19 telah melanda diseluruh dunia, dalam dua tahun terakhir dan mengubah langkah, struktur dan sifat kehidupan bermasyarakat. Penelitian ini  bertujuan untuk mendeteksi covid-19 menggunakan dataset citra chest x-ray yang bersumber dari kaggle.com, yang dibagi menjadi 4 kategori. Metode yang diusulkan yaitu CNN dengan arsitektur MobileNetV2, dengan membagi data train 80% dan data test 20% ukuran citra menjadi 224x224 dan batch size 32. Optimizer menggunakan SGD, lr 0.005, momentum 0.9 serta epoch 20. Hasil penelitian ini dengan capaian nilai presisi untuk kategori covid 0.99, lung opacity 0.98, normal 0.96 dan kategori viral pneumonia mencapai 0.99. Studi selanjutnya dapat menggunakan pengembangan dari model CNN serta dapat mencoba dengan optimizer yang lain.
REAL-TIME STRUCTURAL ANALYSIS BASED ON MACHINE LEARNING FOR CUSTOM PRODUCT DESIGN: A CASE STUDY OF ORTHOPEDIC FIXATOR PRODUCT Aji Digdoyo; Adhitio Satyo Bayangkari Karno; Widi Hastomo; Agita Tunjungsari; Nada Kamilia; Indra Sari Kusuma Wardhana; Nia Yuningsih
J-ICON : Jurnal Komputer dan Informatika Vol 11 No 1 (2023): Maret 2023
Publisher : Universitas Nusa Cendana

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35508/jicon.v11i1.9919

Abstract

Mass customization is related to increasing the balance between the needs of companies that are focused on customers on conditions of production flexibility and efficiency. Product adjustment according to customer needs can increase the company's competitiveness. However, special production processes and adjustments are time consuming and cost inefficient. Parametric product modeling is a fairly popular technique for dealing with this problem. However, it still has challenges related to the high cost of software and a workforce that has special expertise in the field of quality control. In addition, product-specific designs cannot be tested quickly, resulting in a long production time. This study proposes a machine learning (ML) method that aims to obtain a fast time structure to analyze the production of orthopedic fixators. This research process requires a collection of training data with product attributes, physical characteristics, quality, selected ML techniques, and determination of the appropriate set of hyperparameters. Optimization results were obtained using the gradient boosting method with a value of . With these results, the orthopedic fixation device can be used in the case study of developing this machine learning model.
Brain Tumor Classification Using Four Versions of EfficientNet Widi Hastomo; Adhitio Satyo Bayangkari Karno; Dody Arif; Indra Sari Kusuma Wardhana; Nada Kamilia; Rudy Yulianto; Aji Digdoyo; Tri Surawan
Insearch: Information System Research Journal Vol 3, No 01 (2023): Insearch (Information System Research) Journal
Publisher : Fakultas Sains dan Teknologi UIN Imam Bonjol Padang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15548/isrj.v3i01.5810

Abstract

Medical image processing approaches for detecting brain cancers are still primarily done manually, with low accuracy and taking a long period. Furthermore, this task can only be done by professionals with a high degree of medical competence, and the number of experts is obviously restricted in comparison to the large number of patients who need to be treated. With the growth of artificial intelligence and the rapid development of computers in terms of processing speed and storage capacity, it is feasible to assist doctors in classifying the existence of tumors in the head. This study employs four variations of the EfficientNet architecture to train a model on a variety of MRI imaging data. The model version B1 was shown to be the best in this investigation, with 98% accuracy, 99% precision, 95% recall, and 97% f1 score from versions B0 to B3 (4 versions). These results are excellent, but they do not rule out additional study utilizing various forms of design.
Exploratory Data Analysis for Building Energy Meters Using Machine Learning Rudy Yulianto; Sukardi Sukardi; Faqihudin Faqihudin; Meika Syahbana Rusli; Adhitio Satyo Bayangkari Karno; Widi Hastomo; Nia Yuningsih; Nada Kamilia
Journal of Telecommunication Electronics and Control Engineering (JTECE) Vol 5 No 2 (2023): Journal of Telecommunication, Electronics, and Control Engineering (JTECE)
Publisher : LPPM INSTITUT TEKNOLOGI TELKOM PURWOKERTO

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.20895/jtece.v5i2.934

Abstract

The purpose of this research was to apply exploratory data analysis techniques to building energy meters, such as electricity, cold, heat, and steam meters. A thorough understanding of energy usage patterns becomes increasingly vital in an era of growing awareness of energy management and sustainability. Trends, patterns, and anomalies can be identified in building energy meter data using meticulous data exploration approaches, which can give significant insights for increasing energy efficiency. Exploratory data analysis combined with machine learning approaches may be was used to reveal hidden patterns of energy usage and examine the links between relevant factors. The findings of this exploratory data analysis gave vital insights into building energy use trends. Some significant and hidden information that was crucial for understanding energy usage within a certain time frame in each building was discovered via the investigation of the data used in this study. Steam had the highest use, whereas electricity had the lowest. Utilities were more popular before 5 a.m., followed by healthcare, with daytime use hours beginning around 10 a.m., depending on the area. During the working day, the industry needs more energy. Places of worships use more energy on weekends. There was a significant relation between the number of floors and spaces per level of a building and the height meter reading between May and October. There is a significant association between the kind of buildings used for schools, workplaces and high energy use. This study significantly contributed to the management of the energy and sustainability domains. Using exploratory data analysis and machine learning approaches to building energy meters could optimize energy usage, minimize running costs, and enhance overall energy efficiency. This research is still very open to be continued using other methods, to obtain other hidden information.
Comparison of EfficientNet B5-B6 for Detection of 29 Diseases of Fruit Plants Vany Terisia; Widi Hastomo; Adhitio Satyo Bayangkari Karno; Ellya Sestri; Diana Yusuf; Shevty Arbekti Arman; Nada Kamilia
Sainteks Vol 20, No 2 (2023): Oktober
Publisher : Lembaga Penelitian dan Pengabdian Masyarakat (LPPM)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30595/sainteks.v20i2.18691

Abstract

In initiatives to meet food needs and enhance the wellbeing of farmers and society at large, crop production performance is essential. For early attempts to be made for quick handling to prevent crop failure, farmers must be able to readily and quickly receive information in order to detect plant illnesses. In this study, two Convolutional Neural Network (CNN) architectures namely, EfficientNet versions B5 and B6 are used to develop a classification model for plant disease using Deep Learning (DL). The 66,556 visuals in the dataset, which is from Kaggle.com, are used. To create a model, the training method uses 57,067 images data and 3,170 image data for validation. The EfficientNet architecture versions B5 and B6 received very good accuracy scores for the total test results, namely 0.9905 and 0.9927. The model testing phase was carried out through testing phases utilising 3.171 images data. Future analysis can compare CNN architectures and try it with different datasets.