Claim Missing Document
Check
Articles

Found 4 Documents
Search

LONG SHORT-TERM MEMORY FOR PREDICTION OF WAVE HEIGHT AND WIND SPEED USING PROPHET FOR OUTLIERS Baihaqi, Galih Restu; Mulaab
Jurnal Ilmiah Kursor Vol. 12 No. 2 (2023)
Publisher : Universitas Trunojoyo Madura

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.21107/kursor.v12i2.351

Abstract

The reason fishermen lose control is wave height and wind speed. The impact is also felt by all users of the marine sector. This research uses the Long Short Term Memory (LSTM) method because this method has accurate values in the forecasting process with a lot of historical data and uses the Prophet method to detect outliers with Newton interpolation to replace the detected outlier data. The total number of data was 2074 obtained from BMKG Perak Surabaya from January 2020 to November 2022 at four research points, namely north, northeast, east and south points. The test results provide varying error values with MAPE as the model evaluation value. The error value for sea wave height at the north, northeast, east and south points is 13.32 respectively; 13.32; 9.32 and 8.85 with data without interpolation. Meanwhile, the error value in the wind speed data is 14.74; 14.85; 15.14 and 14.52 with a 3rd order Newton interpolation process at the northeast and east points. MAPE values below 20% prove that the LSTM model is good for predicting wave height and wind speed data at four points in Sumenep Regency. The system implementation is made into a web-based application.
Enhancing ResNet with Ghost Weight Normalization For Improved Retina Disease Classification Baihaqi, Galih Restu; Shalsadilla, Shafatyra Reditha; Argaputri, Maulida Khairunisa
IC-ITECHS Vol 5 No 1 (2024): IC-ITECHS
Publisher : LPPM STIKI Malang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32664/ic-itechs.v5i1.1554

Abstract

Retinal disease is a dangerous disease. If left untreated, it can cause blurred vision and even cause permanent blindness. Recently, deep learning approaches are widely used to classify medical diseases. A widely used model to classify medical diseases is ResNet. To train the ResNet model, the data used is data obtained from Kaggle with the name Retinal OCT Images (Optical Coherence Tomography) consisting of 4 classes namely choroidal neovascularization (CNV), DRUSEN, diabetic macular edema (DME), and Normal with a total of 83,600 data. The ResNet base model showed accuracy and f1-score of 92%. Modifying the ResNet Base model with the addition of Ghost Weight Normalization (GWN) which aims to provide more weight normalization opportunities shows an increase in accuracy and f1-score to 94%. GWN can also increase the accuracy of CNN Base from 77% to 81%. This improvement shows that GWN can improve the accuracy of Deep learning models with its weight normalization variation technique. Although the training load and training time when using GWN can increase, the accuracy and f1-score of the ResNet model with GWN of 94% can make the chance of misclassification of retinal diseases smaller.
Accuracy Improvement of Convolutional Neural Network with Ghost Weight Normalization for Pneumonia Classification Baihaqi, Galih Restu; Shalsadilla, Shafatyra Reditha; Argaputri, Maulida Khairunisa
Jurnal Galaksi Vol. 1 No. 3 (2024): Galaksi - Desember 2024
Publisher : Yayasan Sraddha Panca Widya Nusantara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.70103/galaksi.v1i3.35

Abstract

Pneumonia is a critical respiratory condition that requires accurate and timely diagnosis to ensure effective treatment. In this study, we propose the integration of Ghost Weight Normalization (GWN) into a Convolutional Neural Network (CNN) to enhance the accuracy and performance of pneumonia detection. The dataset used was derived from the Kaggle repository, comprising 5,856 chest X-ray images divided into two classes: Normal and Pneumonia. The CNN + GWN model demonstrated improved classification metrics with an accuracy, precision, recall, and F1-score of 95%, outperforming the CNN-Based model, which achieved 92%. While the CNN + GWN model required slightly longer training time and more epochs to achieve its best performance, the trade-off resulted in more robust and reliable predictions. The enhanced performance is attributed to the ability of GWN to normalize weights effectively, providing diverse normalization variations and improving training stability. These results underscore the potential of the CNN + GWN model for reliable pneumonia detection and highlight its capability to address the limitations of conventional CNN architectures.
Implementasi Convolutional Neural Network untuk Klasifikasi Kanker Usus Besar Dengan Normalisasi Ghostweight Baihaqi, Galih Restu; Setiawan, Budi Darma; Muflikhah, Lailil
Jurnal Teknologi Informasi dan Ilmu Komputer Vol 12 No 2: April 2025
Publisher : Fakultas Ilmu Komputer, Universitas Brawijaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.25126/jtiik.2025129432

Abstract

Kanker usus besar merupakan salah satu kanker yang dapat menyebabkan kematian. Jenis kanker ini merupakan kanker peringkat kedua terbanyak pada wanita dan peringkat ketiga terbanyak pada pria. Akhir–akhir ini, pendekatan Deep Learning (DL) banyak digunakan untuk melakukan proses–proses dalam dunia medis. Salah satu metode yang terkenal yaitu Convolutional Neural Network (CNN). CNN tentunya harus memiliki akurasi yang tinggi untuk dapat diterapkan pada kasus ini. Dataset yang digunakan adalah dataset Lung and Colon Cancer Histopathological Images yang berfokus pada kanker ususnya saja. Salah satu cara yang dapat digunakan untuk meningkatkan akurasi pada CNN adalah dengan menormalisasi bobot. Untuk ini, metode yang diusulkan adalah Ghost Weight Normalization (GWN) dengan normalisasi L1 yang terinspirasi dari GhostNet. Metode ini bekerja dengan cara melakukan pembagian bobot utuh menjadi beberapa bagian yang dinamakan GW dan kemudian dinormalisasi untuk setiap GW-nya, lalu digabung kembali menjadi bobot utuh seperti semula. Pendekatan ini terbukti dapat meningkatkan akurasi CNN dengan sangat baik, yaitu mengalami penambahan akurasi sebesar 14% yang semula CNN biasa memperoleh akurasi sebesar 0.8 menjadi 0.94, presisi 0.8 menjadi 0.94 dan F1-score 0.8 menjadi 0.94. GWN juga dapat mengungguli gaya normalisasi biasa, yaitu normalisasi pada bobot tanpa membaginya menjadi GW. Ukuran GW yang efisien adalah 4 dengan perolehan akurasi, persisi, dan f1-score masing-masing 0.94, dengan epoch 8 dan rata – rata untuk waktu proses training-nya pada setiap epoch-nya adalah 259 detik.   Abstract Colon cancer is one of the cancers that can cause death. This type of cancer is the second most common cancer in women and the third most common in men. Lately, Deep Learning (DL) approaches have been widely used to perform processes in the medical world. One of the well-known methods is Convolutional Neural Network (CNN). The method should have high accuracy to be applied in this case. The dataset used is the Lung and Colon Cancer Histopathological Images dataset which focuses on Colon Cancer only. One way that can be used to improve accuracy on CNN is by normalizing the weights. Our proposed method is Ghost Weight Normalization (GWN) with L1 normalization inspired by GhostNet. This method works by dividing the whole weight into several parts called GW and then normalized for each GW, then merged back into the whole weight as before. This approach proved to be able to improve the accuracy of CNN very well, which experienced an increase in accuracy by 14% from the usual CNN accuracy of 0.8 to 0.94, precision 0.8 to 0.94, and f1-score 0.8 to 0.94. GWN can also outperform the usual normalization style, which is normalizing the weights without dividing them into GWs. The efficient GW size is 4 with accuracy, precision, and f1-score of 0.94 each, with 8 epochs and the average training time for each epoch is 259 seconds.