2024年度の研究に関わる業績（研究発表・論文・外部資金・受賞等）です。
The list shows published papers, grants, and awards in 2024-2025 season.

2024年度 / Apr. 2024 〜 Mar. 2025

  • 修士論文 / Master Thesis
    • 佐藤創哉：「枠情報を用いた書類の項目間の関係性抽出」
    • 中込集杜：「深層学習を用いた非侵襲嚥下機能評価の可視化」
    • 堀田慎：「膝軟骨信号を用いた深層学習による変形性膝関節症の判別」※修士論文優秀発表賞
    • 矢島英明：「大規模視覚言語モデルを用いた書類画像における文字の検出と認識」※修士論文優秀発表賞
  • 卒業論文 / Graduate Thesis
    • 植松凌大：「音声合成モデルを用いた超高齢者向け音声認識モデルの改善」
    • 阪口直紀：「大規模言語モデルを用いた質疑応答タスクにおける生成テキスト抑制手法」
    • 藤田光基：「文字認識モデル訓練のための複数文字入力に基づく手書き文字生成」
    • 武藤直輝：「外国人留学生のための人工知能技術を用いた日本語授業理解支援システム」※卒業論文優秀発表賞
  • 雑誌論文 (Journal): 8 papers
    • Ahmad Hakimi Ahmad Sa’ahiry, Abdul Halim Ismail, Masahiro Toyoura, Latifah Munirah Kamaruddin, Mohd Sani Mohamad Hashim, Muhamad Safwan Muhamad Azmi, Hiromitsu Nishizaki, Xiaoyang Mao, “Device diversity in crowdsourced WiFi fingerprinting database for autonomous mobile robot”, International Journal of Advanced Robotic Systems, Vol.21, No.6, pp.1-16, Nov./2024, DOI: 10.1177/17298806241297, IF: 2.3
      Abstract: Positioning and navigation of mobile robot is the main feature for the trajectory or motion of the mobile robot. Conventional mobile robot positioning and navigation system relies heavily on fusion of multiple costly sensors, which does not promote mass production. This paper aim is to use readily and available technologies which is WiFi due to its reliability as it is pre-deployed, and it exist in most of the building. The system used are based on indoor positioning system (IPS) by using a crowdsourced fingerprinting method. This seeks to improve crowdsourced fingerprinting database performance by solving the issue of the device diversity or heterogeneity of difference devices. To cope with the crowdsourced fingerprinting database as the location estimation method for the robot application, deep neural network (DNN) is employed. The proposed method namely ratio and ranged-based (RRB) shows an improvement of 60% increments by implementing the pre-processing technique of the raw data before feeding it to the DNN. The comparison between other method shows that RRB is better in term of accuracy in three validation techniques, which are root mean square error (RMSE), distance error and accuracy between true and estimate position. This improvement effectively could facilitate the actual positioning system utilizing the WiFi infrastructure for the mobile robot in very near future.
    • Noraini Azmi, Latifah Munirah Kamarudin, Ahmad Shakaff Ali Yeon, Ammar Zakaria, Syed Muhammad Mamduh Syed Zakaria, Hiromitsu Nishizaki, Latifah Mohamed, Xiaoyang Mao, Mohd Hafiz Fazalul Rahiman, “Integration of dual band radio waves and ensemble-based approach for rice moisture content determination and localisation”, Journal of Stored Products Research, Vol.108, Pages 102399, Sep/2024, DOI: 10.1016/j.jspr.2024.102399, IF: 2.7
      Abstract: Maintaining optimal moisture content in grain storage is critical to ensuring adequate supply throughout the year, but it presents a significant challenge. Current moisture measurement methods often necessitate sophisticated and costly equipment. This paper introduces an approach employing real-time rice moisture content determination and detection of spoilage (specifically wet spots) within a storage facility achieved through the utilisation of radio waves operating at 2.4 GHz and 868 MHz, along with an ensemble-based machine learning algorithm. Experimental samples spanning from 12% to 30% moisture levels were collected, then subjected to pre-processing, and subsequently employed to train the Ensemble-based Rice Moisture Content and Localisation (eRMCL) algorithm. The eRMCL produced an effective prediction of both rice moisture content and the localisation of wet spots within the grain storage unit. The results show that compared to support vector machine, random forest, and machine learning methods, the eRMCL algorithm had the best performance metrics, with an accuracy of 94.8% in predicting the moisture content and location of spoilage in storage. The measurement of moisture content and the identification of wet spots in rice storage using the dual frequency wave approach were found to be more accurate than with a single frequency band. Thus, the dual frequency band is a novel method for the determination of the moisture content of stored rice and the localisation of the spoilage area.
    • Chuo Hiang Heng, Masahiro Toyoura, Chee Siang Leow, Hiromitsu Nishizaki, “Analysis of Classroom Processes Based on Deep Learning With Video and Audio Features,” IEEE Access, vol. 12, pp. 110705-110712, 2024, DOI: 10.1109/ACCESS.2024.3434742, IF:3.9
      Abstract: An active learning-type class is a class in which students take the initiative. In order to improve active learning-type classes, attempts have been made to review the content after the class using video taken by the lecturer, but this is burdensome because it takes a long time to review the video. Although methods have been proposed for estimating the classroom process at each time, there is still room for improvement. In this paper, we propose a method for estimating the classroom process at each time with higher accuracy than conventional methods. The proposed method uses deep learning to improve the accuracy of the conventional method, which only uses classical SVM. We also used an ablation study to find an appropriate combination of input modalities. Furthermore, we introduced ensemble LSTM to handle data with different modalities. The proposed method achieved the highest accuracy of 98.8% and the lowest accuracy of 64.4% in estimating classroom activities, with an average classification accuracy of 80.1%.
    • Hui Fern Soon, Amiza Amir, Hiromitsu Nishizaki, Nik Adilah Hanin Zahri, Latifah Munirah Kamarudin, “An FPA-Optimized XGBoost Stacking for Multi-Class Imbalanced Network Attack Detection”, International Journal of Advanced Computer Science and Applications, Vol.15, No.7, pp. 1380-1390, 2024. DOI:10.14569/IJACSA.2024.01507134, IF: 2.1
      Abstract: Network anomaly detection systems face challenges with imbalanced datasets, particularly in classifying underrepresented attack types. This study proposes a novel framework for improving F1-scores in multi-class imbalanced network attack detection using the UNSW-NB15 dataset, without resorting to resampling techniques. Our approach integrates Flower Pollination Algorithm-based hyperparameter tuning with an ensemble of XGBoost classifiers in a stacking configuration. Experimental results show that our FPA-XGBoost-Stacking model significantly outperforms individual XGBoost classifiers and existing ensemble models. The model achieved a higher overall weighted F1-score compare to the individual XGBoost classifier and Thockchom et al.’s heterogeneous stacking ensemble. Our approach demonstrated remarkable effectiveness across various levels of class imbalance, for example Analysis and Backdoor which is highly underrepresented classes, and DoS which is moderately underrepresented class. This research contributes to more effective network security systems by offering a solution for imbalanced classification without resampling techniques’ drawbacks. It demonstrates that homogeneous stacking with XGBoost can outperform heterogeneous approaches for skewed class distributions. Future work will extend this approach to other cybersecurity datasets and explore its applicability in real-time network environments.
    • Hwai Ing Soon, Azian Azamimi Abdullah, Hiromitsu Nishizaki, Mohd Yusoff Mashor, Latifah Munirah Kamarudin, Zeti-Azura Mohamed-Hussein, Zeehaida Mohamed, Wei Chern Ang, “Optimizing hybrid neural networks for precise COVID-19 mRNA vaccine degradation prediction”, International Journal of Advanced and Applied Sciences, Vol. 11, No. 7, pp.87-100, July/2024, DOI: 10.21833/ijaas.2024.07.011, IF: 0.4
      Abstract: Conventional hybrid models often miss an essential factor that can lead to less effective performance: intrinsic sequence dependence when combining various neural network (NN) architectures. This study addresses this issue by highlighting the importance of sequence hybridization in NN architecture integration, aiming to improve model effectiveness. It combines NN layers—dense, long short-term memory (LSTM), and gated recurrent unit (GRU)—using the Keras Sequential API for defining the architecture. To provide better context, bidirectional LSTM (BiLSTM) and bidirectional GRU (BiGRU) replace their unidirectional counterparts, enhancing the models through bidirectional structures. Out of 25 NN models tested, 18 four-layer hybrid NN models consist of one-quarter dense layer and the rest BiLSTM and BiGRU layers. These hybrid NN models undergo supervised learning regression analysis, with mean column-wise root mean square error (MCRMSE) as the performance metric. The results show that each hybrid NN model produces unique outcomes based on its specific hybrid sequence. The Hybrid_LGSS model performs better than existing three-layer BiLSTM networks in predictive accuracy and shows lower overfitting (MCRMSEs of 0.0749 and 0.0767 for training and validation, respectively). This indicates that the optimal hybridization sequence is crucial for achieving a balance between performance and simplicity. In summary, this research could help vaccinologists develop better mRNA vaccines and provide data analysts with new insights for improvement.
    • Hwai Ing Soon, Azian Azamimi Abdullah, Hiromitsu Nishizaki, Mohd Yusoff Mashor, Latifah Munirah Kamarudin, Zeti-Azura Mohamed-Hussein, Zeehaida Mohamed, Wei Chern Ang “COVID-19 mRNA vaccine degradation rate prediction using artificial intelligence techniques: A narrative review”, International Journal of Advanced and Applied Sciences, Vol. 11, No. 6, pp.215-228, June/2024, DOI: 10.21833/ijaas.2024.06.023, IF: 0.4
      Abstract: As diseases become more common, the use of mRNA (messenger ribonucleic acid) vaccines is becoming more important. These vaccines can be developed quickly and have a low risk of side effects. However, they are sensitive to environmental conditions, which means they need careful storage and transport, creating challenges in distributing them. Testing the stability of an mRNA vaccine requires a lot of work and time, as it needs many lab tests. Artificial Intelligence (AI) offers a new solution by using the genetic information in RNA sequences to predict how quickly these vaccines might break down. This approach helps address potential shortages of vaccines by avoiding some of the challenges with vaccine distribution. The COVID-19 pandemic has greatly sped up the use of AI in this area. This change is significant because using AI to predict and improve the stability of mRNA vaccines was not well explored before the pandemic. This paper reviews recent studies that use AI to study mRNA vaccines during the COVID-19 pandemic. It points out that the main issue with these vaccines is how long they can be stored before they are no longer effective due to their sensitivity to environmental conditions. By looking at these studies, the paper not only shows how AI and vaccine research are coming together but also points out opportunities for more research. The goal of this review is to outline effective methods to improve the use of mRNA vaccines and encourage more scientific research and development in this field. This is an important step in improving how we deal with pandemics.
    • Tasneem Sofri, Allan Melvin Andrew, Hasliza A Rahim, Hiromitsu Nishizaki, Latifah Munirah Kamarudin, Peng Wen Wong, Ping Jack Soh, “Enhancing Predictive Models for Assessing 5G Exposure Effects on Human Health and Cognition through Supervised Machine Learning: A Multi-Stage Feature Selection Approach”, Przeglad Elektrotechniczny, Vol.2024, No.6, pp.122-128, June/2024. DOI: 10.15199/48.2024.06.23
      No prior reviews have focused on any comprehensively examine the effects of 5G exposure (700 MHz to 30 GHz) on human health and cognition using supervised Machine Learning (ML). This novel research combined the Multi-Stage Feature Selection (MSFS) and hybrid features for classification machine learning model. The approach which includes the use of MSFS, yielded better results in terms of accuracy, precision, F1- score, sensitivity, and specificity when contrasted with the approach that did not incorporate MSFS with accuracy more than 0.95 for both datasets.
  • 国際会議論文 (Reviewed conference papers): 12 papers
    • Ren Fujimoti, Chee Siang Leow, Hideaki Yajima, Koji Makino, Xiaoyang Mao, Hiromitsu Nishizaki, “Development of QR Code-Guided Autonomous Navigation System for Grape Cultivation Robot in Overhead Trellis Vineyard”, Proceedings of the 2025 IEEE International Conference on Industrial Technology, pp.1-6, Mar. 2025, Presented on 27th/Mar/2025 in Wuhan, China
    • Ze Li, Yuke Lin, Yao Tian Hongbin Suo, Pengyuan Zhang, Yanzhen Ren, Zexin Cai, Hiromitsu Nishizaki, and Ming Li, “The Database and Benchmark for the Source Speaker Tracing Challenge 2024”, Proceedings of the IEEE Spoken Language Technology Workshop 2024, pp.-, Macau, Dec 2 – 5, Dec./2024, Presented on 5th/Dec/2024 in Macau
    • So Watanabe, Chee Siang Leow, Junichi Hoshino, Takehito Utsuro, Hiromitsu Nishizaki, “Assessment and Improvement of Customer Service Speech with Multiple Large Language Models”, Proceedings of the 2024 APSIPA Annual Summit and Conference, pp.1-6, Macau, Dec 3 – 6, Dec./2024, Presented on 5th/Dec/2024 in Macau, PDF
    • Leow Chee Siang, Tsutomu Tanzawa, Bong Tze Yaw, Koji Makino, Kazuyoshi Ishida, Hiromitsu Nishizaki, “Development of a Fruit Theft Reporting System Using a Compact Microcontroller with Deep Learning Based on Suspicious Sounds”, Proceedings of the 50th Annual Conference of the IEEE Industrial Electronics Society (IECON 2024), pp.1-6, Nov./2024. Presented on 4th/Nov/2024 in Cicago, USA
    • Ryosuke Shimazu, Chee Siang Leow, Prawit Buayai, Koji Makino, Xiayang Mao, Hiromitsu Nishizaki, “High Quality Color Estimation of Shine Muscat Grape Using Vision Transformer”, Proceedings of the 23rd International Conference on Cyberworlds (Cyberworlds 2024), pp.195-202, Oct/2024. DOI:10.1109/CW64301.2024.00028, Presented on 31st/October in Kofu, Yamanashi, Japan ※Best Paper Award
    • Hwai Ing Soon, Azian Azamimi Abdullah, Hiromitsu Nishizaki, Latifah Munirah Kamarudin, “Prediction Model with Penalized Hyperparameter Optimization for mRNA Vaccine Degradation Based on Tetra-nitrogenous-base Analysis”, Proceedings of the 23rd International Conference on Cyberworlds (Cyberworlds 2024), pp.249-256, Oct/2024. DOI:10.1109/CW64301.2024.00035, Presented on 31st/October in Kofu, Yamanashi, Japan
    • Muhammad Rasydan Mazlan, Abdul Syafiq Abdull Sukor, Abdul Hamid Adom, Latifah Munirah Kamarudin, Hiromitsu Nishizaki, Norasmadi Abdul Rahim, “Analysis of Acute Stress Reaction with EEG using MI-mRMR EEG Channel Selection and Ensemble Learning with Majority Voting”, Proceedings of the 23rd International Conference on Cyberworlds (Cyberworlds 2024), pp.272-279, Oct/2024. DOI:10.1109/CW64301.2024.00038, Presented on 30st/October in Kofu, Yamanashi, Japan
    • Jae Hyun Lee, Ye-Seul Jang, Hiromitsu Nishizaki, Won-Du Chang, “The dynamic load-induced bioelectric potentials of knee joint”, Proceedings of the 23rd International Conference on Cyberworlds (Cyberworlds 2024), pp.358-359, Oct/2024. DOI:10.1109/CW64301.2024.00072, Presented on 30st/October in Kofu, Yamanashi, Japan
    • Makoto Hotta, Chee Siang Leow, Norihide Kitaoka, Hiromitsu Nishizaki, “Evaluation of Speech Translation Subtitles Generated by ASR With Unnecessary Word Detection”, Proceedings of the 2024 IEEE 13th Global Conference on Consumer Electronics (GCCE), pp.835-839, Oct/2024, Presented on 31st/Oct in Kitakyusyu, Fukuoka, DOI: 10.1109/GCCE62371.2024.10760522
    • Hideaki Yajima, Chee Siang Leow, Hiromitsu Nishizaki, “Text Detection and Style Classification From Images Using Vision Transformer and Transformer Decoder”, Proceedings of the 2024 IEEE 13th Global Conference on Consumer Electronics (GCCE), pp.632-636, Oct/2024, Presented on 30th/Oct in Kitakyusyu, Fukuoka, DOI: 10.1109/GCCE62371.2024.10761042
    • Kazuyoshi Ishida, Chee Siang Leow, Tsutomu Tanzawa, Tze Yaw Bong, Hiromitsu Nishizaki, Koji Makino, “Evaluation of LoRa-Based Long-Range Communication in a Fruit Theft Prevention Device”, Proceedings of the 2024 IEEE 13th Global Conference on Consumer Electronics (GCCE), pp.291-294, Oct/2024, Presented on 29th/Oct in Kitakyusyu, Fukuoka, DOI: 10.1109/GCCE62371.2024.10760408
    • Shuto Nakagomi, Yutaka Suzuki, Masayuki Morisawa, Hiromitsu Nishizaki, Muramatsu Hideaki, Motoki Arakawa, “Development of GUI Application for Multimodal Analysis for Evaluation of Swallowing Function”, Proceedings of the 2024 IEEE 13th Global Conference on Consumer Electronics (GCCE), pp.974-978, Oct/2024, Presented on 30th/Oct in Kitakyusyu, Fukuoka, DOI: 10.1109/GCCE62371.2024.10760652
  • 口頭発表 (Domestic conference, not reviewed) 6 papers
    • 西尾直樹，レオ チーシャン，西崎博光，”特徴点マッチングと寸法を活用した2D-CAD図面検索”，情報処理学会第87回全国大会講演論文集，2U-09，pp.-，2025年3月（2025/3/13発表，吹田市）
    • 遠藤陽季，矢島英明，レオ チーシャン，丹沢勉，牧野浩二，石田和義，西崎博光，”農作物盗難防止のための小型エッジデバイスで動作する音分類モデル”，情報処理学会第87回全国大会講演論文集，5V-01，pp.-，2025年3月（2025/3/14発表，吹田市）
    • 藤本 蓮，レオ チーシャン，矢島英明，牧野浩二，茅 暁陽，西崎博光，”ブドウ栽培支援ロボットのためのQRコード誘導型自律移動システムの開発”，情報処理学会第87回全国大会講演論文集，5J-02，pp.-，2025年3月（2025/3/14発表，吹田市）
    • 佐藤創哉，矢島英明，レオ チーシャン，西崎博光，”書類画像における枠の関連性を考慮した項目間の関係性抽出モデル”，情報処理学会第87回全国大会講演論文集，4ZM-02，pp.-，2025年3月（2025/3/14発表，吹田市）
    • 渡辺蒼，レオ チーシャン，西崎博光，星野准一，宇津呂 武仁，”複数の大規模言語モデルを用いた円卓会議による接客評価”，人工知能学会第38回全国大会論文集，No. 4Xin2-100，pp.1-4，2024.5（2024/5/31発表，浜松市）
    • 長谷川遼，銭本友樹，宇津呂武仁，西崎博光，吉岡真治，神門典子，”大規模言語モデルによる自由記述アンケート自動集約のための疑似訓練事例生成”，人工知能学会第38回全国大会論文集，No. 1J4-OS-10b-03，pp.1-4，2024.5（2024/5/28発表，浜松市）
  • 外部資金新規採択分 (Grant, only new acceptance)
    • 「多機能ロボット開発と栽培体系革新によるシャインマスカット高効率栽培の実現」，次世代スマート農業技術の開発・改良・実用化，生物系特定産業技術研究支援センター，令和6年9月〜令和9年3月，茅暁陽（代表），西崎博光・レオチーシャン（分担）
    • 「高精度AI-OCRモデル構築のための多様な手書き文字画像の自動生成モデリング」，科学研究費支援事業，研究活動スタート支援，令和6年8月〜令和8年3月，レオチーシャン（代表）
    • 「改良型音AI駆動の果実盗難検知通報システムの開発と大規模検証補助事業」，2024年度機械振興補助事業振興事業補助，公益財団法人JKA，令和6年4月〜令和7年3月、西崎博光・レオチーシャン（分担）
  • 書籍・雑誌記事 (Books, Magazines)
  • 学外授業・セミナー
    • 山梨県DX・情報政策推進統括官主催（山梨大学協力）「AI・データ利活用スペシャリスト養成講座」
      • 西崎博光：”Pythonスキルアップ講座①〜データ操作編〜” 2024年10月2日実施
      • 西崎博光：”Pythonスキルアップ講座②〜API活用編〜” 2024年10月9日実施
      • 西崎博光：”生成AI入門セミナー①〜文章生成AI編〜” 2024年11月6日実施
      • 西崎博光：”生成AI入門セミナー②〜画像生成AI編〜” 2024年11月13日実施
    • 山梨県DX・情報政策推進統括官主催（山梨大学協力）「YAMANASHI AIハッカソン2024：生成AIを活用してアプリを作ってみよう！」実施日：2024年11月24日，2024年12月1日，2024年12月8日，担当講師：レオ チーシャン
  • 表彰・報道等
    • 『ブドウ摘粒を支援 試作ロボット実演』，2024年7月9日，山梨日日新聞朝刊
    • 『ぶどう摘粒作業時の粒数を自動で測定するスマートフォンアプリ「粒羅(tsubura)」に、収穫適期を判定する新機能を搭載』，2024年8月5日，山梨大学プレスリリース
    • 『シャイン収穫AIが判定 山梨大アプリ新機能色合いで見極め』，2024年8月20日，山梨日日新聞朝刊