Faculty Profiles - SHIGEI Noritaka

Papers - SHIGEI Noritaka

Division display 21 - 40 of about 193 ／ All the affair displays >>

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 . 2023 IEEE International Magnetic Conference - Short Papers, INTERMAG Short Papers 2023 - Proceedings 1 - 2 2023.5

Language：Japanese Publisher：2023 IEEE International Magnetic Conference - Short Papers, INTERMAG Short Papers 2023 - Proceedings

Optimization of magnetic components design, such as power inductors and transformers, is most needed to improve the performance of future power electronics. However, power electronics designers are faced with the problem of not having sufficient magnetic component models available for their designs. In this paper, we propose a method to construct a unique nonlinear magnetic component model using parametric circuit simulation and deep learning.

DOI： 10.1109/INTERMAGShortPapers58606.2023.10228481

Scopus

Miyajima Hirofumi, Shigei Noritaka, Miyajima Hiromi, Shiratori Norio . Scalability improvement of simplified, secure distributed processing with decomposition data . Nonlinear Theory and Its Applications, IEICE14 ( 2 ) 140 - 151 2023.4Reviewed

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

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.

DOI： 10.1587/nolta.14.140

YAMAGUCHI Yuki, SHIGEI Noritaka, MIYAZAKI Masanobu, ISHIZUKA Yoichi, ABE Shinichi, NISHINO Tomoya, MIYAJIMA Hiromi . Classification of Renal Function Trends in Time Series Data Using Machine Learning . Journal of Japan Society for Fuzzy Theory and Intelligent Informatics35 ( 1 ) 511 - 516 2023.2Reviewed

Authorship：Corresponding author Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Japan Society for Fuzzy Theory and Intelligent Informatics

As a preventive measure for chronic kidney disease requiring dialysis, some municipal governments and medical associations make comments to family doctors and examinees based on the results of health checkups. However, the judgment and preparation of comments are mostly done manually by physicians and public health nurses and require much labor. In this study, we examine the trend determination of kidney function level using machine learning as part of the study to realize the automation of this task. We propose a preprocessing method for time-series data dealing with different numbers of checkups and different duration of checkups. As a machine learning model, we propose ensemble learning methods using gradient boosting decision trees. The effectiveness of the proposed methods is demonstrated in evaluation using about 3,000 cases of specific health checkup data.

DOI： 10.3156/jsoft.35.1_511

MIYAJIMA Hirofumi, SHIGEI Noritaka, MIYAJIMA Hiromi, SHIRATORI Norio . Computational Complexity Reduction of Secure BP Learning Algorithm with Divided Data . Journal of Japan Society for Fuzzy Theory and Intelligent Informatics35 ( 1 ) 506 - 510 2023.2Reviewed

Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Japan Society for Fuzzy Theory and Intelligent Informatics

Many studies have been conducted on how to perform learning while maintaining data security. One of them is the secure distributed processing with decomposition data. The feature of this method is that it achieves high confidentiality by decomposing data and parameters in machine learning and performing distributed processing. On the other hand, realization of machine learning by distributed and integrated processing of decomposed data and parameters leads to an increase in computational complexity and degradation of computational accuracy. In this paper, we propose an improved BP learning method that suppresses the increase in computational complexity with the increase in the number of servers, and demonstrate its effectiveness.

DOI： 10.3156/jsoft.35.1_506

Yoshiki Higo, Noritaka Shigei, Hiromi Miyajima, Hirofumi Miyajima . Computation Time Reduction and Accuracy Improvement of Secure Multiparty Computation for Back Propagation by Introducing Communication Timeout . Proc. of International Symposium on Artificial Life and Robotics 479 - 484 2023.1Reviewed

Hirofumi Miyajima; Noritaka Shigei; Hiromi Miyajima; Norio Shiratori . Secure Distributed Processing of NG with Updatable Decomposition Data and Parameters . 2022 International Conference on Networking and Network Applications (NaNA) 354 - 360 2022.12Reviewed

Shimasaki Ryota, Shigei Noritaka, Miyajima Hirofumi, Nakamura Yoshihiro, Miyajima Hiromi . Improvement of Self-Driving Model Car for Indoor Patrol Task . International Journal of Innovative Computing, Information and Control18 ( 06 ) 1829 - 1839 2022.12Reviewed

Miyajima H., Shigei N., Miyajima H., Shiratori N. . Secure Distributed Processing of NG with Updatable Decomposition Data and Parameters . Proceedings - 2022 International Conference on Networking and Network Applications, NaNA 2022 354 - 360 2022.12Reviewed

Language：Japanese Publishing type：Research paper (international conference proceedings) Publisher：Proceedings - 2022 International Conference on Networking and Network Applications, NaNA 2022

Machine learning using distributed data, such as federative learning (FL) and secure multiparty computation (SMC), is demanded to achieve both utility and confidentiality when using confidential data. There is a trade-off between utility and confidentiality, and in general, SMC can offer better confidentiality than FL and better utility than homomorphic encryption. In machine learning with SMC, confidentiality is improved by decomposing individual data and parameters into multiple pieces, storing each piece on each server, and learning without restoring the data or parameters themselves. However, once the conventional methods randomly decompose data and parameters, the decomposition remains permanently fixed. The fixed decomposition is considered undesirable because it gives malicious attackers more opportunities to attack the data and model. In this paper, we propose a secure distributed processing of neural gas (NG), which is one of unsupervised machine learning. In addition to the decomposition of data, the proposed method can update the decomposition of parameters during learning. Each server can independently update the decomposition of both data and parameters in the proposed method, and the data and parameters are never restored during learning. Our simulation result shows that it can achieve the same level of learning accuracy as the conventional methods.

DOI： 10.1109/NaNA56854.2022.00067

Scopus

MIYAJIMA Hirofumi, SHIGEI Noritaka, MIYAJIMA Hiromi, SHIRATORI Norio . Reducing the Computational Complexity of Machine Learning Methods Through Secure Distributed Processing . Proceedings of the Annual Conference of Biomedical Fuzzy Systems Association35 ( 0 ) I-1 2022.12

Language：Japanese Publishing type：Research paper (conference, symposium, etc.) 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). AI analysis of big data will bring effective information that meets the needs of individuals and companies to the real society more quickly. On the other hand, to build a safe and secure society, it is important to develop AI analysis methods that protect the privacy of big data in cyberspace. However, there is no known method that satisfies both data confidentiality and usability of the learning method at a high level. Therefore, the authors have proposed a distributed AI method using secret decomposition data. This method has higher confidentiality and usability than conventional methods, but the increase in computational complexity due to distributed processing is a problem. In this paper, we propose a new learning method to solve this problem. In particular, we apply this method to the Neural-Gas(NG) algorithm, which is unsupervised learning, and show its effectiveness.

DOI： 10.24466/pacbfsa.35.0_i-1

Yamaguchi Yuki, Shigei Noritaka, Miyazaki Masanobu, Ishizuka Yoichi, Abe Shinichi, Nishino Tomoya, Miyajima Hiromi . Classification of Kidney Function Level from Time Series Data Using Machine Learning . Proceedings of the Fuzzy System Symposium38 ( 0 ) 741 - 745 2022.9

Authorship：Corresponding author Language：Japanese Publishing type：Research paper (conference, symposium, etc.) Publisher：Japan Society for Fuzzy Theory and Intelligent Informatics

In this study, we aim to construct effective classification models of renal function level using machine learning, which can be used by physicians in the diagnosis of renal function. We propose a data preprocessing method for the constructed classification models to deal with different numbers and different periods of medical checkups. As machine learning models, we use gradient boosting decision tree models and propose several ensemble learning methods. The effectiveness of the proposed methods is demonstrated by simulation results with about 3,000 cases of medical checkup data.

DOI： 10.14864/fss.38.0_741

Miyajima Hirofumi, Shigei Noritaka, Miyajima Hiromi, Shiratori Norio . The improved BP learning algorithm with divided data . Proceedings of the Fuzzy System Symposium38 ( 0 ) 281 - 286 2022.9

Language：Japanese Publishing type：Research paper (conference, symposium, etc.) Publisher：Japan Society for Fuzzy Theory and Intelligent Informatics

Many studies for privacy preserving data mining have been done. Secure Multiparty Computation (SMC) has been introduced for privacy preserving data. SMC was applied to some statistical computation methods. In previous papers, we proposed synchronous Back Propagation (BP) learning for SMC. In this paper, we propose the improved BP learning method with small number of parameters, even if the number of data and weight divisions is large.

DOI： 10.14864/fss.38.0_281

Miyajima H., Shigei N., Miyajima H., Shiratori N. . Securely Distributed Computation with Divided Data and Parameters for Hybrid Particle SwarmOptimization . IAENG International Journal of Applied Mathematics52 ( 3 ) 2022.8Reviewed

Language：English Publishing type：Research paper (scientific journal) Publisher：IAENG International Journal of Applied Mathematics

Machine learning (ML) on cloud and edge systems is widely used to analyze big data and address complex problems. On the other hand, many privacy issues have arisen due to inadequate data security controls. Therefore, studies on secure machine learning on cloud and edge systems have attracted much attention. One of such studies is Federated Learning (FL), in which data is distributed on multiple servers to achieve machine learning through distributed processing. We have proposed a method to realize learning by distributed processing on multiple servers by dividing the learning data and parameters into multiple pieces in advance and distributing these pieces to each server. In previous papers, we have proposed a learning method using local search based on the Steepest Descent Method (SDM) such as Back Propagation (BP) and Neural Gas (NG). In this paper, we propose a learning method that combines a global and local search for solutions for secure distributed processing computation. In particular, we propose a secret distributed processing method for Hybrid Particle Swarm Optimization (HPSO) which is one of the global and local search methods. Its effectiveness is demonstrated by numerical simulations.

Scopus

Kazuya ARIMURA, Noritaka SHIGEI, Chihiro MORITA, Yoichi ISHIZUKA, Hiromi MIYAJIMA . Rust appearance evaluation methods for weathering steel by using bagging CNN classifier and multiple patch images . Journal of Japan Society for Fuzzy Theory and Intelligent Informatics 2022.5Reviewed

Kohei OGURA, Noritaka SHIGEI, Hiromi MIYAJIMA . Robust person tracking by pan-tilt camera with low-cost single board computer . ICIC Express Letters16 ( 4 ) 2022.4Reviewed International journal

Hirofumi Miyajima, Noritaka Shigei, Hiromi Miyajima, Norio Shiratori . Secure Learning Systems using Vertically Partitioned Data with IoT . IAENG International Journal of Computer Science48 2022.3Reviewed International journal

西結人, 重井徳貴 . Pix2pixを用いた急傾斜地崩壊危険箇所の推定 . 火の国情報シンポジウム2022 1 - 8 2022.3

Kenshiro Yamanaka, Noritaka Shigei, Hirofumi Miyajima, Hiromi Miyajima . Face image recognition resistant to posture fluctuations using intermediate template vectors and UMAP . 760 - 764 2022.1Reviewed

Hirofumi Miyajima, Noritaka Shigei, Hiromi Miyajima, Norio Shiratori . The generalized BP method of secret distributed processing with divided data . 691 - 696 2022.1Reviewed

宮島洋文, 重井徳貴, 宮島廣美, 白鳥則郎 . 秘匿分解データを用いた新しい機械学習 . 研究報告マルチメディア通信と分散処理（DPS）2021DPS189 ( 11 ) 1 - 8 2021.12

千々和航, 重井徳貴 . サーモグラフィーと深層学習を用いた非接触入力装置における背景情報を用いた識別精度改善 . 日本知能情報ファジィ学会九州支部学術講演会予稿集 64 - 68 2021.11

　 PREV - NEXT 　

<SHIGEI Noritaka>

Papers - SHIGEI Noritaka