Faculty Profiles - SHIGEI Noritaka

Papers - SHIGEI Noritaka

Division display 1 - 20 of about 193 ／ All the affair displays >>

Li Shijia, Noritaka Shigei . A Preliminary Study on Calibration Method of Low-Cost MOX-type CO2 Sensors for Indoor Occupancy Estimation . Proceedings of the Fuzzy System Symposium 44 - 47 2025.9

Tsubasa HIDAKA, Kanta KUBO, Kan TANABE, Rara DEGUCHI, Haruka ETO, Kenji BABA, Naoki KUROSHIMA, Masumi WADA, Mashiho MUKAIDA, Takao OHTSUKA, Satoshi ONO, Noritaka SHIGEI . A Data Augmentation Method for Temporal Action Segmentation in Laparoscopic Surgery Videos . Proceedings of the Fuzzy System Symposium 472 - 475 2025.9

Miyajima H., Shigei N., Miyajima H., Shiratori N. . BP Methods for Securely Distributed System with Communication Failures . Lecture Notes in Networks and Systems1553 326 - 340 2025.9

Hirofumi Miyajima, Noritaka Shigei, Hiromi Miyajima and Norio Shiratori . Neural Gas method using autonomous and secure distributed processing with decomposed data . Nonlinear Theory and Its Applications, IEICE16 ( 3 ) 377 - 389 2025.7Reviewed

Language：English Publishing type：Research paper (scientific journal) Publisher：The Institute of Electronics, Information and Communication Engineers

<p>Much research on secure and safe AI analysis methods for users has been conducted. They are 1) secure multiparty computation (SMC), 2) quasi-homomorphic encryption, 3) federated learning (FL), and so on. It is known that both utility and confidentiality are essential for machine learning using confidential data. However, there is a trade-off between them. In the previous paper, we proposed a learning method for secure distributed processing using decomposition data and parameters. The characteristic feature of this method is to achieve high confidentiality through distributed processing with decomposed data and parameters. On the other hand, the realization of machine learning through the distribution and integration of decomposition data leads to an increase in computational complexity and a deterioration of computational accuracy. In particular, while increasing the number of servers improves confidentiality, this problem becomes more pronounced. In this paper, we propose scalability improvement of simplified, secure distributed processing methods of BP and NG methods to suppress the computational complexity associated with an increase in the number of servers and demonstrate its effectiveness through a numerical simulation.</p>

DOI： 10.1587/nolta.16.377

Miyajima H., Shigei N., Miyajima H., Shiratori N. . Toward the development of learning methods with distributed processing using securely divided data . Computers and Electrical Engineering123 2025.4Reviewed International journal

Authorship：Corresponding author Language：English Publishing type：Research paper (scientific journal) Publisher：Computers and Electrical Engineering

To pave the way to a super-smart society, artificial intelligence (AI) methods are being developed to discover and analyze necessary information instantly from cyberspace and utilize it in physical space. However, privacy protection is necessary for AI to process big data in cyberspace. From the viewpoint of developing safe and secure machine learning methods, research on (1) homomorphic cryptography, (2) differential privacy, (3) secure multiparty computation, and (4) federated learning is underway. The goal of these studies is to develop useful learning methods while maintaining data privacy. We propose a method to address the trade-off between security and usability in machine learning. This method balances usability and data confidentiality by using decomposed data to achieve secure distributed processing. However, such methods using distributed processing increase computational and communication overhead as the number of servers increases. To address this problem, we propose a method to control the computational complexity as the number of servers increases. On the basis of these studies, this study first systematically addresses the construction of secure distributed processing methods with decomposed data. A comprehensive approach is essential to advance the field and allow these methods to be effectively applied to different domains. On the basis of these methods, we propose back-propagation and neural gas learning methods with reduced computational and communication requirements. We then apply the proposed methods to numerical simulations of class classification and clustering problems and show that accuracy comparable to that of conventional models can be achieved with 1/Q computational and communication complexity for distributed models with Q servers.

DOI： 10.1016/j.compeleceng.2025.110160

Scopus

Other Link： https://www.sciencedirect.com/science/article/pii/S004579062500103X

Nagatani K., Shigei N., Morita C., Ishizuka Y., Miyajima H. . CNN AND GAN BASED WEATHERING STEEL APPEARANCE EVALUATION IMPROVED BY IMAGE SIZE AND LOSS FUNCTION . International Journal of Innovative Computing, Information and Control21 ( 2 ) 433 - 445 2025.4Reviewed International journal

Authorship：Corresponding author Language：English Publishing type：Research paper (scientific journal) Publisher：International Journal of Innovative Computing, Information and Control

Weathering steel is a steel material that can maintain its performance outdoors for a long period of time by forming protective rust. Periodic inspection of the rust condition is necessary to maintain this steel in good condition, and CNN-based methods for evaluating the appearance of rust have been studied. One of the problems in improving the classification accuracy of the CNN is how to prepare the rust images for training. The objectives of this paper are 1) to consider rust patch image size and 2) to improve the diversity of the generated images in rust image generation using GAN. From 1), we show that the accuracy can be improved by using large cropping images and shrinking the image when training the model. From 2), we propose SSIM-target loss, which is an improvement of the conventional SSIM loss, and show that it improves the CNN Score. Using methods 1) and 2), we achieved an F1 score of 90.3% for the three-class classification of the rust images, compared to the baseline method’s F1 score of 86.8%.

DOI： 10.24507/ijicic.21.02.433

Scopus

Other Link： http://www.ijicic.org/ijicic-210211.pdf

Ryuo Kawabata, Kotaro Nagata, Yukari Nakamaru, Kentaro Yasui, Chihiro Morita, and Noritaka Shigei . Detection of Concrete Cracks Using YOLOv8 and Estimation of Crack Width . Proceedings of International Symposium on Collaborative Informatics T2-4-1 - T2-4-8 2025.3Reviewed

Tsubasa HIDAKA, Kanta KUBO, Kan TANABE, Rara DEGUCHI, Kenji BABA, Naoki KUROSHIMA, Masumi WADA, Mashiho MUKAIDA, Takao OHTSUKA, Satoshi ONO, Noritaka SHIGEI . Proposal of Ensemble Method for Action Segmentation in Laparoscopic Surgery Videos . Journal of Japan Society for Fuzzy Theory and Intelligent Informatics37 ( 1 ) 571 - 576 2025.2Reviewed

Hirofumi Miyajima, Noritaka Shigei, Hiromi Miyajima, Norio Shiratori . BP Learning Method using Secure Distributed Processing with Decomposition Data and Parameters Divided into Updated and Non-updated Portions . Proc. of International Symposium on Artificial Life and Robotics 625 - 630 2025.1Reviewed

Shuhei Hamashima, Noritaka Shigei, Hiromi Miyajima, Masanobu Miyazaki, Yoichi Ishizuka, Shinichi Abe, Tomoya Nishino . Construction of Regression Model of eGFR by Data Partitioning Based on Trend of Time Series Data . Proc. of International Symposium on Artificial Life and Robotics 563 - 568 2025.1Reviewed

Rara Deguchi, Kan Tanabe, Tsubasa Hidaka, Kenji Baba, Naoki Kuroshima, Masumi Wada, Mashiho Mukaida, Takao Ohtsuka, Noritaka Shigei, Satoshi Ono . A Preliminary Study on Surgical Skill Classification from Laparoscopic Surgery Videos Using Multi-Instance Learning . Proc. of International Symposium on Artificial Life and Robotics 274 - 277 2025.1Reviewed

Nagatani K., Shigei N., Morita C., Ishizuka Y., Miyajima H. . Improvement of GAN-based Training Image Generation for CNN-based rust Evaluation of Weathering Steel . ACM International Conference Proceeding Series 40 - 47 2024.12Reviewed International journal

Authorship：Corresponding author Language：English Publishing type：Research paper (international conference proceedings) Publisher：ACM International Conference Proceeding Series

Weathering steel is a steel material that produces a protective rust layer on the surface of the steel that suppresses the propagation of corrosion. Periodic inspection of the rust condition is necessary to maintain this steel in good condition, and CNN-based methods for evaluating the appearance of rust have been studied. One of the problems in improving the classification accuracy of the CNN is how to prepare the rust images for training. Since it is difficult to collect rust images in the real world, it has been studied to train CNN with rust images generated by GANs. Conventional GANs have the problem that when a large number of images are generated, the generated images are biased toward similar ones. The scope of this paper is to improve the accuracy of CNN using GAN-generated images in classification tasks of rust conditions of weathering steel. The specific objectives of this study are to improve the diversity of images generated by GANs and thereby improve the accuracy of CNNs. To improve image diversity in GAN-based image generation, we propose to use Wasserstein GAN with Gradient Penalty (WGAN-GP), in addition to Conditional GAN (CGAN), and to introduce SSIM (Structural Similarity) and LBP (Local Binary Pattern) as loss functions, in addition to L1 and L2 losses. Then, the images generated by the proposed five types of GANs are applied to the training of CNNs. Comparison between the proposed GANs and the conventional CGAN shows that 1) the proposed GANs can reduce the SSIM similarity of the generated images by up to 1/4, and 2) in the classification of three classes of rust images, the use of the generated images of the proposed GANs can improve the F1 score by up to about 4%pt. Overall analysis shows that WGAN-GP-based is better than CGAN-based and that the newly introduced SSIM, LBP, L1 and L2 losses improve both similarity and classification accuracy.

DOI： 10.1145/3688268.3688275

Scopus

Hirofumi Miyajima, Noritaka Shigei, Hiromi Miyajima, Norio Shiratori . Neural Gas Method using Autonomous and Secure Distributed Processing with Decomposed Data . Proc. of International Symposium on Nonlinear Theory and Its Applications 137 - 140 2024.12Reviewed

Hirofumi Miyajima, Noritaka Shigei, Hiromi Miyajima, Norio Shiratori . Autonomous and Securely Distributed Processing of BP Method Using Decomposed Data . International Journal of Biomedical Soft Computing and Human Sciences29 ( 2 ) 37 - 44 2024.10Reviewed

日高翼, 久保莞太, 田辺寛, 出口楽々, 馬場研二, 黒島直樹, 和田真澄, 向田眞志保, 大塚隆生, 小野智司, 重井徳貴 . 腹腔鏡下手術映像におけるアクションセグメンテーションの分類モデルの最適化とアンサンブル手法の検討 . 日本知能情報ファジィ学会ファジィシステムシンポジウム講演論文集40 ( 0 ) 648 - 653 2024.9

Miwa Keigo, Shigei Noritaka, Sugimoto Satoshi, Ishizuka Yoichi, Miyajima Hiromi . Estimation of Sediment Disaster Prone Areas from Map Images Using Mask R-CNN . ICIC Express Letters, Part B: Applications15 ( 06 ) 575 - 584 2024.6Reviewed International journal

Authorship：Corresponding author Language：English Publishing type：Research paper (scientific journal) Publisher：ICIC International

In Japan, where sediment disasters such as debris flow, landslides, and slope failures frequently occur, for disaster prevention purposes, the government publishes areas at risk of sediment disasters as Sediment Disaster Prone Areas. Since this requires time and effort for experts to conduct field surveys, attempts are being made to estimate the collapse hazard areas from elevation maps using deep learning. However, the conventional method of estimating collapse hazard areas pixel by pixel has the problem that it is difficult to identify the area clearly. To solve these problems, we propose using instance segmentation and object detection in computer vision techniques. This study investigates effective methods for estimating areas at risk of sediment disasters using a Mask R-CNN model that simultaneously processes segmentation and object detection. Specifically, we develop 1) a method for generating mask images, 2) a learning method combining Red Relief Image Map images, aerial photographs, standard maps, and land use maps as input images, and 3) a method for overlapping prediction. Numerical experiments conducted under multiple conditions evaluate the accuracy and the predicted images and demonstrate the effectiveness of the developed methods.

DOI： 10.24507/icicelb.15.06.575

Scopus

Other Link： http://www.icicelb.org/ellb/contents/2024/6/elb-15-06-04.pdf

Yamaguchi Yuki, Shigei Noritaka, Miyazaki Masanobu, Ishizuka Yoichi, Abe Shinichi, Nishino Tomoya, Miyajima Hiromi . Development of Assistance System to Diagnose Changes in Kidney Function Using Machine Learning . ICIC Express Letters18 ( 06 ) 615 2024.6Reviewed

MIYAJIMA Hirofumi, SHIGEI Noritaka, MIYAJIMA Hiromi, SHIRATORI Norio . Neural Gas and K-Means Methods Improving the Computational Complexity for Secure Distributed Processing . International Journal of Biomedical Soft Computing and Human Sciences: the official journal of the Biomedical Fuzzy Systems Association28 ( 1 ) 15 - 22 2023.11Reviewed

Language：English Publisher：Biomedical Fuzzy Systems Association

To realize a super-smart society, it is necessary to aim for advanced integration of cyberspace and physical space (real society). Artificial Intelligence (AI) analysis of big data will bring effective information that meets the needs of individuals and companies to real society more quickly. On the other hand, to build a safe and secure society, it is important to develop AI methods that protect the privacy of big data in cyberspace. However, there is a little-known method that satisfies both data confidentiality and utilization of the learning method. Therefore, the authors proposed a learning method for secure distributed processing using decomposition data. This method has higher confidentiality and utilization than the conventional method, but the increase in computational complexity due to distributed processing is a problem. In the previous paper, the authors proposed the Back Propagation (BP) method to solve this problem. In this paper, we apply this method to the Neural Gas (NG) and k-means methods for secure distributed processing, which are unsupervised learning, and show its effectiveness.

DOI： 10.24466/ijbschs.28.1_15

Motomatsu T., Koga T., Shigei N., Yamaguchi M., Itagaki A., Ishizuka Y. . A Data-Driven Inductor Modeling Technique Using Parametric Circuit Simulation and Deep Learning . IEEE Transactions on Magnetics59 ( 11 ) 1 - 5 2023.11Reviewed International journal

Miyajima H., Shigei N., Miyajima H., Shiratori N. . Neural Gas and k-means Methods with Reduced Communication Costs for Secure Distributed Processing . Lecture Notes in Engineering and Computer Science2245 19 - 24 2023.7Reviewed

　 PREV - NEXT 　

<SHIGEI Noritaka>

Papers - SHIGEI Noritaka