Researchers Database

TAKATA Masami

FacultyFaculty Division of Human Life and Environmental Sciences Research Group of Information and Communication Technology for Life
PositionLecturer
Last Updated :2022/10/08

researchmap

Profile and Settings

  • Name (Japanese)

    Takata
  • Name (Kana)

    Masami

Research Areas

  • Informatics, Software

Research Experience

  • May 2013, 奈良女子大学・講師
  • Apr. 2006, Apr. 2013, 奈良女子大学・助教
  • Apr. 2004, Mar. 2006, PRESTO, JST ・ Resercher

Education

  • Apr. 2001, Mar. 2004, Nara Women's University, Graduate School, Doctral Research Course in Human Culture, Division of Integrated Science

Teaching Experience

  • pasage (47B), Nara Women's University, 99 Apr. 2019
  • pasage (47A), Nara Women's University, 99 Apr. 2019
  • 生活環境学における研究倫理, Nara Women's University, 99 Apr. 2019

Association Memberships

  • IPSJ
  • 情報処理学会\nハイパフォーマンスコンピューティング研究会
  • 情報処理学会\n数理モデルと問題解決研究会
  • 情報処理学会\n若手研究者の会

Ⅱ.研究活動実績

Published Papers

  • Refereed, Advances in Parallel & Distributed Processing, and Applications, Translating Early-modern Written Style into Current Colloquial Style in Hoji Shinbun, Honoka Nishikawa; Yuki Takemoto; Sayaka Iida; Yu Ishikawa; Masami Takata; Kaoru Ueda; Kazuki Joe, 2023, International conference proceedings
  • Refereed, Advances in Parallel & Distributed Processing, and Applications, Application of Deep Metric Learning to Early-modern Japanese Printed Character Recognition, Norie Koiso; Yuki Takemoto; Sayaka Iida; Yu Ishikawa; Masami Takata; Kazuki Joe, 2023, International conference proceedings
  • Refereed, Advances in Parallel & Distributed Processing, and Applications, Improved Fitness Functions for 3D Packing Problem, Shigeka Nakajima; Masami Takata, 2023, International conference proceedings
  • Refereed, Advances in Parallel & Distributed Processing, and Applications, Proposal on Scheduling Method for Sightseeing with Unlimited Ride Tickets, Shiori Higano; Reina Isono; Masami Takata, 2023, International conference proceedings
  • Refereed, Advances in Parallel & Distributed Processing, and Applications, On a Fast Computation of the Condition L-curve in the LSMR Method for Solving Ill-conditioned Problems using the DQDS Method, Itsuki Kuboi; Rika Tanaka; Shin-ya Ozawa; Yohsuke Hosoda; Masami Takata; Kinji Kimura; Yoshimasa Nakamura, 2023, International conference proceedings
  • Refereed, 情報処理学会論文誌数理モデル化と応用(TOM), Method of Acquiring Low-frequency Character Types for Specific Early-Modern Japanese Printed Book Publishers, Yuki Takemoto; Yu Ishikawa; Masami Takata; Kazuki Joe, 26 Jul. 2022, 15, 3, 71, 89, Scientific journal
  • Refereed, 研究報告数理モデル化と問題解決(MPS), On a fast computation of the condition L-curve in the LSMR method for solving ill-conditioned problems using the DQDS method, 久保井 五貴; 田中 利佳; 小澤 伸也; 細田 陽介; 髙田 雅美; 木村 欣司; 中村 佳正, Jul. 2022, 2022-MPS-139, 3, 1, 4, Symposium
  • Refereed, 研究報告数理モデル化と問題解決(MPS), Proposal on Scheduling Method for Sightseeing with Unlimited Ride Tickets, Shiori Higano; Reina Isono; Masami Takata, Jul. 2022, 2022-MPS-139, 4, 1, 4, Symposium
  • Refereed, 研究報告数理モデル化と問題解決(MPS), Application of Deep Metric Learning to Early-modern Japanese Printed Character Recognition, Norie Koiso; Yuki Takemoto; Sayaka Iida; Yu Ishikawa; Masami Takata; Kazuki Joe, Jul. 2022, 2022-MPS-139, 10, 1, 4, Symposium
  • Refereed, 研究報告数理モデル化と問題解決(MPS), 3次元箱詰め問題にための適応度関数の改良, 中島 滋香; 髙田 雅美, Jul. 2022, 2022-MPS-139, 6, 1, 4, Symposium
  • Refereed, IPSJ Transactions on Mathematical Modeling and Its Applications, Accelerating the Numerical Computation of Positive Roots of Polynomials Using Suitable Combination of Lower Bounds, Masami Takata; Takuto Akiyama; Sho Araki; Hiroyuki Ishigami; Kinji Kimura; Yoshimasa Nakamura, Jan. 2022, 15, 1, 18, 30, Scientific journal
  • Refereed, Springer Nature - Advances in Parallel & Distributed Processing, and Applications, Generating an Estimated Font Set for Early-Modern Japanese Printed Character Recognition, Yuki Takemoto; Yu Ishikawa; Masami Takata; Kazuki Joe, 2022, International conference proceedings
  • Refereed, Springer Nature - Advances in Parallel & Distributed Processing, and Applications, Implementation of Singular Value Decomposition for Complex Matrices Using the Two-sided Jacobi Method, Takahiro Miyamae; Masami Takata; Kinji Kimura; Yoshimasa Nakamura, 2022, International conference proceedings
  • Refereed, Springer Nature - Advances in Parallel & Distributed Processing, and Applications, Proposal of an Essential Oil Blending Method using Q-learning, Kyoka Nakano; Masami Takata, 2022, International conference proceedings
  • Refereed, Advances in Parallel & Distributed Processing, and Applications, Improvement of Island GA using Multiple Fitness Functions, Shigeka Nakajima; Masami Takata, Nov. 2021, 725, 736, International conference proceedings
  • Refereed, Advances in Parallel & Distributed Processing, and Applications, On an Implementation of the One-Sided Jacobi Method with High Accuracy, Masami Takata; Sho Araki; Kinji Kimura; Yoshimasa Nakamura, Nov. 2021, 713, 724, International conference proceedings
  • Refereed, Advances in Parallel & Distributed Processing, and Applications, Application of the Orthogonal QD Algorithm with Shift to Singular Value Decomposition for Large Sparse Matrices, Hiroki Tanaka; Taiki Kimura; Tetsuaki Matsunawa; Shoji Mimotogi; Masami Takata; Kinji Kimura; Yoshimasa Nakamura, Nov. 2021, 697, 712, International conference proceedings
  • Refereed, Advances in Parallel & Distributed Processing, and Applications, Crawling Low Appearance Frequency Characters Images for Early-Modern Japanese Printed Character Recognition, Nanami Fujisaki; Yu Ishikawa; Masami Takata; Kazuki Joe, Nov. 2021, 683, 695, International conference proceedings
  • Refereed, IPSJ Transactions on Mathematical Modeling and Its Applications, On an Implementation of the One-Sided Jacobi Method With High Accuracy, Masami Takata; Sho Araki; Takahiro Miyamae; Kinji Kimura; Yoshimasa Nakamura, Aug. 2021, 14, 3, 68, 75, Scientific journal
  • Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 近代書籍フォントの推定生成, 竹本有紀; 石川由羽; 高田雅美; 城和貴, Jul. 2021, 2021-MPS-134, 13, 1, 4, Symposium
  • Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 両側ヤコビ法を用いた複素数行列に対する特異値分解の実装法, 宮前隆広; 高田雅美; 木村欣司; 中村佳正, Jul. 2021, 2021-MPS-134, 8, 1, 4, Symposium
  • Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), Q-learningを用いた精油ブレンド法の提案, 中野杏香; 高田雅美, Jul. 2021, 2021-MPS-134, 7, 1, 3, Symposium
  • Refereed, IPSJ Transactions on Mathematical Modeling and Its Applications, On an Implementation of the Two-Sided Jacobi Method, Sho Araki; Masana Aoki; Masami Takata; Kinji Kimura; Yoshimasa Nakamura, Jan. 2021, 14, 1, 12, 20, Scientific journal
  • Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 複数の適応度関数を用いた島GAの改良, 中島 滋香; 高田 雅美, Jul. 2020, 2020-MPS-129, 5, 1, 4
  • Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 高精度な特異値分解を実現する片側Jacobi法の実装方法, 髙田 雅美; 荒木 翔; 木村 欣司; 中村 佳正, Jul. 2020, 2020-MPS-129, 4, 1, 4
  • Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 大規模行列の特異値分解へのOQDS法の適用, 田中 博基; 木村 泰己; 松縄 哲明; 三本木 省次; 髙田 雅美; 木村 欣司; 中村 佳正, Jul. 2020, 2020-MPS-129, 3, 1, 4
  • Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 近代書籍用文字認識のための低出現頻度文字クローリング, 藤﨑 菜々美; 竹本 有紀; 石川 由羽; 髙田 雅美; 城 和貴, Jul. 2020, 2020-MPS-129, 2, 1, 4
  • Refereed, In Proceedings of 2019 International Conference on Parallel and Distributed Processing Techniques and Applications, Layout Analysis using Semantic Segmentation for Imperial Meeting Minutes, Sayaka Iida; Yuki Takemoto; Yu Ishikawa; Masami Takata; Kazuki Joe, Jul. 2019, 135, 141
  • Refereed, In Proceedings of 2019 International Conference on Parallel and Distributed Processing Techniques and Applications, A Discrete Three-wave System of Kahan-Hirota-Kimura Type and the QRT Mapping, Yuko Takae; Masami Takata; Kinji Kimura; Yoshimasa Nakamura, Jul. 2019, 142, 148
  • Refereed, In Proceedings of 2019 International Conference on Parallel and Distributed Processing Techniques and Applications, Improvement of the Thick-Restart Lanczos Method in Single Precision Floating Point Arithmetic using Givens Rotations, Masana Aoki; Masami Takata; Kinji Kimura; Yoshimasa Nakamura, Jul. 2019, 149, 155
  • Refereed, In Proceedings of 2019 International Conference on Parallel and Distributed Processing Techniques and Applications, On an Implementation of Two-Sided Jacobi Method, Sho Araki; Masami Takata; Kinji Kimura; Yoshimasa Nakamura, Jul. 2019, 156, 162
  • Refereed, In Proceedings of 2019 International Conference on Parallel and Distributed Processing Techniques and Applications, Structure of Neural Network Automatically Generating Fonts for Early-Modern Japanese Printed Books, Yuki Takemoto; Yu Ishikawa; Masami Takata; Kazuki Joe, Jul. 2019, 182, 188
  • Refereed, In Proceedings of 2019 International Conference on Parallel and Distributed Processing Techniques and Applications, Applying CNNs to Early-Modern Printed Japanese Character Recognition, Suzuka Yasunami; Norie Koiso; Yuki Takemoto; Yu Ishikawa; Masami Takata; Kazuki Joe, Jul. 2019, 189, 195, International conference proceedings
  • Refereed, IPSJ Transactions on Mathematical Modeling and Its Applications, Improvement of the Augmented Implicitly Restarted Lanczos Bidiagonalization Method in Single Precision Floating Point Arithmetic, Yuya Ishida; Masami Takata; Kinji Kimura; Yoshimasa Nakamura, Dec. 2018, 11, 3, 19-25
  • Refereed, In Proceedings of 2018 International Conference on Parallel and Distributed Processing Techniques and Applications, Automatic Font Generation For Early-Modern Japanese Printed Books, TAKATA Masami; Yuki Takemoto; Yu Ishikawa; Masami Takata; Kazuki Joe, Jul. 2018, 326-332
  • Refereed, In Proceedings of 2018 International Conference on Parallel and Distributed Processing Techniques and Applications, Fast Computation Method of Column Space by using the DQDS Method and the OQDS Method, TAKATA Masami; Sho Araki; Hiroki Tanaka; Masami Takata; Kinji Kimura; Yoshimasa Nakamura, Jul. 2018, 333-339
  • Refereed, In Proceedings of 2018 International Conference on Parallel and Distributed Processing Techniques and Applications, Improvement of Computing Partial Singular Value Decomposition for Dense Matrix using Thick-restart-Lanczos Method, TAKATA Masami; Masana Aoki; Masami Takata; Kinji Kimura; Yoshimasa Nakamura, Jul. 2018, 340-346
  • Refereed, 情報処理学会論文誌. 数理モデル化と応用, Implementation of Computing Partial Singular Value Decomposition for Principal Component Analysis Using ARPACK, TAKATA Masami, Mar. 2018, 11, 1, 37-44
  • Refereed, In Proceedings of 2017 International Conference on Parallel and Distributed Processing Techniques and Applications, Validation of EEG Authentication Accuracy with Electrode Slippage, TAKATA Masami; Yu Ishikawa; Kaori Nishibata; Masami Takata; Hiroyasu Kamo; Kazuki Joe, Jul. 2017, 302-308
  • Refereed, In Proceedings of 2017 International Conference on Parallel and Distributed Processing Techniques and Applications, Hand shape recognition using the sEMG of the upper extremity, TAKATA Masami; Mayuko Iwamoto; Yu Ishikawa; Masami Takata; Kazuki Joe, Jul. 2017, 295-301
  • Refereed, In Proceedings of 2017 International Conference on Parallel and Distributed Processing Techniques and Applications, Early-Modern Printed Character Recognition using Ensemble Learning, TAKATA Masami; Kaori Fujimoto; Yu Ishikawa; Masami Takata; Kazuki Joe, Jul. 2017, 288-294
  • Refereed, In Proceedings of 2017 International Conference on Parallel and Distributed Processing Techniques and Applications, An Improvement of Augmented Implicitly Restarted Lanczos Bidiagonalization Method, TAKATA Masami; Yuya Ishida; Masami Takata; Kinji Kimura; Yoshimasa Nakamura, Jul. 2017, 281-287
  • Refereed, 情報処理学会論文誌. 数理モデル化と応用, 脳波から求めた複数特徴量の組合せによる生体認証, TAKATA Masami, Mar. 2017, 10, 1, 23-32
  • Refereed, 情報処理学会論文誌. 数理モデル化と応用, Performance Evaluation of Golub-Kahan-Lanczos Algorithm with Reorthogonalization by Classical Gram-Schmidt Algorithm and OpenMP, TAKATA Masami; Masami Takata; Hiroyuki Ishigami; Kinji Kimura; Yuki Fujii; Hiroki Tanaka; Yoshimasa Nakamura, The Golub-Kahan-Lanczos algorithm with reorthogonalization (GKLR algorithm) is an algorithm for computing a subset of singular triplets of large-scale sparse matrices. The reorthogonalization tends to become a bottleneck of the execution time, as the iteration number of the GKLR algorithm increases. In this paper, OpenMP-based parallel implementation of the classical Gram-Schmidt algorithm with reorthogonalization (OMP-CGS2 algorithm) is introduced. The OMP-CGS2 algorithm has the advantage of data reusability and is expected to achieve higher performance of the reorthogonalization computations on shared-memory multi-core processors with large caches than the conventional reorthogonalization algorithms. Numerical experiments on shared-memory multi-core processors show that the OMP-CGS2 algorithm accelerates the GKLR algorithm more effectively for computing a subset of singular triplets of some matrices than the conventional reorthogonalization algorithms. In addition, we discuss the cache utilization in the OMP-CGS2 algorithm and a condition that the OMP-CGS2 algorithm achieves higher performance than the CGS2 algorithm.The Golub-Kahan-Lanczos algorithm with reorthogonalization (GKLR algorithm) is an algorithm for computing a subset of singular triplets of large-scale sparse matrices. The reorthogonalization tends to become a bottleneck of the execution time, as the iteration number of the GKLR algorithm increases. In this paper, OpenMP-based parallel implementation of the classical Gram-Schmidt algorithm with reorthogonalization (OMP-CGS2 algorithm) is introduced. The OMP-CGS2 algorithm has the advantage of data reusability and is expected to achieve higher performance of the reorthogonalization computations on shared-memory multi-core processors with large caches than the conventional reorthogonalization algorithms. Numerical experiments on shared-memory multi-core processors show that the OMP-CGS2 algorithm accelerates the GKLR algorithm more effectively for computing a subset of singular triplets of some matrices than the conventional reorthogonalization algorithms. In addition, we discuss the cache utilization in the OMP-CGS2 algorithm and a condition that the OMP-CGS2 algorithm achieves higher performance than the CGS2 algorithm., Aug. 2016, 9, 2, 1-8, 8
  • Refereed, 情報処理学会論文誌. 数理モデル化と応用, 近代書籍を対象とした多フォント漢字認識, TAKATA Masami, 国立国会図書館では,所蔵する明治から昭和前期の近代書籍を近代デジタルライブラリーとしてWeb上でページごとの画像データとして公開しているが,全文検索を行うことができない.そのため,自動テキスト化が望まれている.すでに我々は近代書籍に適したオフライン多フォント漢字認識手法を提案しているが,対象とした漢字は256種であり,確かな有効性を示すには至っていない.これは,近代書籍特有のルビが除去できなかったことが主な原因であるが,これに関してもすでにルビ除去手法を提案しており,ようやく学習データの大幅な増加が期待できるようになった.そこで本論文では,2634種の漢字を学習データとした識別器を生成し,92%の認識率を得た.さらに,実際に近代デジタルライブラリーで公開されている書籍に生成した識別器を適用し,有効性を検証した.結果,認識率は平均90%を超えており,良好な結果が得られた.In the web site of National Diet Library, the digital library from the Meiji era is open to the public. Since the early-modern Japanese printed books are given as image data, namely, full-text search is not available, automatic conversion to the text is needed. Already we have proposed the Multi-Fonts Kanji character recognition method for Early-Modern Printed Books. The proposed method was intended for only 256 types of Kanji characters, it failed to show a certain validity because there is few types of Chinese characters. The reason is because training data were not collected by the failure of removal of a ruby. In order to solve the problem, we have proposed a ruby removal method. As the result, it is possible to increase the training data. In this paper, by using the 2634 type of Kanji characters, we demonstrate the effectiveness of the proposed method., Aug. 2016, 9, 2, 33-40, 40
  • Refereed, In Proceedings of 2016 International Conference on Parallel and Distributed Processing Techniques and Applications, Implementation of Computing Singular Pairs for Large Scale Matrices using ARPACK, TAKATA Masami; Masami Takata; Sho Araki; Kinji Kimura; Yuki Fujii; Yoshimasa Nakamura, Jul. 2016, 349-355
  • Refereed, In Proceedings of 2016 International Conference on Parallel and Distributed Processing Techniques and Applications, Biometric Authentication based on Multi-feature Combination using EEG, TAKATA Masami; Yu Ishikawa; Kaori Nishibata; Masami Takata; Hiroyasu Kamo; Kazuki Joe, Jul. 2016, 394-400
  • Refereed, In Proceedings of 2016 International Conference on Parallel and Distributed Processing Techniques and Applications, Correlation of Proximity Voluntary Muscles EMG and EEG, TAKATA Masami; Hitomi Oigawa; Yu Ishikawa; Umeda Tomohiro; Masami Takata; Kazuki Joe, Jul. 2016, 401-407
  • Refereed, In Proceedings of 2016 International Conference on Parallel and Distributed Processing Techniques and Applications, Comparison of Feature Extraction Methods for Early-Modern Japanese Printed Character Recognition, TAKATA Masami; Kazumi Kosaka; Kaori Fujimoto; Yu Ishikawa; Masami Takata; Kazuki Joe, Jul. 2016, 408-414
  • Refereed, In Proceedings of 2015 International Conference on Parallel and Distributed Processing Techniques and Applications, A Personal Classification Method Using Spatial Information of Multi-channel EEG, TAKATA Masami; Yu Ishikawa; Chinami Yoshida; Masami Takata; Hiroyasu Kamo; Kazuki Joe, Jul. 2015, I, 229-235
  • Refereed, In Proceedings of 2015 International Conference on Parallel and Distributed Processing Techniques and Applications, Performance Evaluation of Golub-Kahan-Lanczos Algorithm with Reorthogonalization by Classical Gram-Schmidt Algorithm and OpenMP, TAKATA Masami; Masami Takata; Hiroyuki Ishigami; Kinji Kimura; Yuki Fujii; Hiroki Tanaka; Yoshimasa Nakamura, Jul. 2015, I, 243-249
  • Refereed, In Proceedings of 2015 International Conference on Parallel and Distributed Processing Techniques and Applications, A Music Composition Model with Genetic Programming -A Case Study of Chord Progression and Bassline-, TAKATA Masami; Kanae Kunimatsu; Yu Ishikawa; Masami Takata; Kazuki Joe, Jul. 2015, I, 256-262
  • Refereed, In Proceedings of 2015 International Conference on Parallel and Distributed Processing Techniques and Applications, An Effective and Interactive Training Data Collection Method for Early-Modern Japanese Printed Character Recognition, TAKATA Masami; Kazumi Kosaka; Taeka Awazu; Yu Ishikawa; Masami Takata; Kazuki Joe, Jul. 2015, I, 276-282
  • Refereed, In Proceedings of 2015 International Conference on Parallel and Distributed Processing Techniques and Applications, Visualization of Sensory Weight for Shouldering Randoseru, TAKATA Masami; Hitomi Oigawa; Yu Ishikawa; Masami Takata; Kazuki Joe, Jul. 2015, I, 297-303
  • Refereed, 情報処理学会論文誌. 数理モデル化と応用, 活字データの分類を用いた進化計算による近代書籍からのルビ除去, TAKATA Masami, Mar. 2015, 8, 1, 72-79
  • Refereed, In Proceedings of 2014 International Conference on Parallel and Distributed Processing Techniques and Applications, Information Processing Society of Japan (IPSJ), Accelerating the Numerical Computation of Positive Roots of Polynomials using Improved Bounds, TAKATA Masami; Kinji Kimura; Takuto Akiyama; Hiroyuki Ishigami; Masami Takata; Yoshimasa Nakamura, The continued fraction method for isolating the positive roots of a univariate polynomial equation is based on Vincent's theorem, which computes all of the real roots of polynomial equations. In this paper, we propose two new lower bounds which accelerate the fraction method. The two proposed bounds are derived from a theorem stated by Akritas et al., and use different pairing strategies for the coefficients of the target polynomial equations from the bounds proposed by Akritas et al. Numerical experiments show that the proposed lower bounds are more effective than existing bounds for some special polynomial equations and random polynomial equations, and are competitive with them for other special polynomial equations., Jul. 2014, II, 6, 201-207, 4
  • Refereed, In Proceedings of 2014 International Conference on Parallel and Distributed Processing Techniques and Applications, Validation of EEG Personal Authentication with Multi-channels and Multi-tasks, TAKATA Masami; Yu Ishikawa; Chinami Yoshida; Masami Takata; Kazuki Joe, Jul. 2014, II, 182-188
  • Refereed, 7th international joint conference on BIOmedical engineering Systems and TEChnologies, Biometrics Authentication using Another Feature of Heartbeat Waveform, TAKATA Masami; Rhoko Nomura; Tomohiro Umeda; Naoko Yoshii; Masami Takata; Kazuki Joe, Mar. 2014, CD
  • Refereed, 3rd International Conference on Pattern Recognition Applications and Methods, A Multi-fonts Kanji Character Recognition Method for Early-modern Japanese Printed Books with Ruby Characters, TAKATA Masami; Taeka Awazu; Manami Fukuo; Masami Takata; Kazuki Joe, Mar. 2014, CD
  • Refereed, SEMANTIC TECHNOLOGY, SPRINGER-VERLAG BERLIN, Ontology Construction Support for Specialized Books, Yuki Eguchi; Yuri Iwakata; Minami Kawasaki; Masami Takata; Kazuki Joe, In this paper, we present a support system for ontology construction just based on a given specialized book with lowest possible cost. The system tries to combine minimum hand and mostly automatic constructions. It extracts required information for the ontology design from the specialized book with presenting yes-no selections to an expert of the specialized book. The ontology construction is performed automatically just using the answers of the expert. The constructed ontology is reliably and highly technical since it is constructed on the basis of the specialized book. In addition, since user operations are restricted only to yes-no selection, any expert can make use of our system without any special knowledge about ontology., 2014, 8388, 159, 174, International conference proceedings
  • Refereed, 2014 7TH BIOMEDICAL ENGINEERING INTERNATIONAL CONFERENCE (BMEICON), IEEE, Biometrics Authentication Based on Chaotic Heartbeat waveform, Ryoko Nomura; Yu Ishikawa; Tomohiro Umeda; Masami Takata; Hiroyasu Kamo; Kazuki Joe, In this paper, we propose a personal authentication method using heartbeat waveforms to enhance the security in wireless communication. In the heartbeat waveforms of a human being, it is known that there are discriminative characteristic features and the mimicking is very difficult. Therefore, their application to personal authentication has been studied. Existing method performs personal authentication using a chaos indicator of ECG. In this paper, we propose an authentication method that extends the index in order to increase the accuracy of authentication. To validate that our personal authentication method is applicable, we perform some experiments to show that our method provides better authentication than existing one., 2014, 1-5, International conference proceedings
  • Refereed, 情報処理学会論文誌 数理モデル化と応用, Generating Matrices Having Large Condition Numbers for Performance Evaluation of Singular Value Decomposition Algorithms, TAKATA Masami, 本稿では,特異値分解を評価するために,条件数の大きなテスト行列の作成法を提案する.我々が対象とする条件数は,以下の2種類の定義によるものである.1つ目は,連立1次方程式を解く際の困難さの指標として知られる最大特異値と最小特異値の比による定義である.2つ目は,特異値分解の数値計算の困難さを表す特異値の近接度による条件数の定義である.1つ目の条件数の大きなテスト行列の作成法では,行列のべき乗を利用するもので,密行列を作成することが可能である.一方,2つ目の作成法では,近接固有値を持つglued Wilkinson行列の特異値版ともいえるもので,2重対角行列のみが作成可能である.提案する2種類の作成法の目的は異なるため,それぞれに意義がある.これらの作成法によって作成されるテスト行列を用いて,LAPACK 3.4.2に含まれているいくつかの特異値分解アルゴリズムを評価する.In this paper, we propose new generating algorithms for matrices with large condition number to evaluate singular value decomposition. We target two types of definitions of condition numbers. The first is a rario of the maximal and minimal singular values and means the intractableness to solve simultaneous linear equation. The second definition uses adjacent amount in each singular value and indicates the difficulty of numerical singular value decomposition. The first algorithm, which is calculated to a power, can generate dense test matrices. On the other hand, the second algorithm can generate only a bidiagonal test matrix which is a extension of the glued Wilkinson matrix having very closed eigenvalues. Since targets in the proposed algorithms are different, it is important to generate test matrices in two types of condition numbers. By using resulting test matrices which are generated by the proposed algorithms, some singular value decomposition routines in LAPACK version 3.4.2 are evaluated., Dec. 2013, 6, 3, 75-86, 86
  • Refereed, 情報処理学会論文誌 コンピューティングシステム, New Shift Strategy in the dqds Algorithm and the m2dLVs Algorithm for Singular Values, TAKATA Masami, Sep. 2013, 6, 3, 94-107
  • Refereed, 情報処理学会論文誌 数理モデル化と応用, α波構成要素の切出しと解析, TAKATA Masami, Aug. 2013, 6, 2, 93-103
  • Refereed, 情報処理学会論文誌 数理モデル化と応用, Ruby Removal Filters Using Genetic Programming for Early-modern Japanese Printed Books, TAKATA Masami, 国立国会図書館では,所蔵する明治から昭和前期の近代書籍を近代デジタルライブラリとしてWEB上でページごとの画像データとして公開しているが,文書内容での検索を行うことができない.そのため,自動でのテキストデータ化が望まれている.その際,問題となっているのがルビである.現在のルビを直線的に除去する技術は,規格に沿った現在の書籍を対象としたものであるため,現在の書籍とは違う特性を持つ近代書籍には適用できない.そこで,本研究では,遺伝的プログラミングを用いて,曲線的に出版者・時代ごとの専用ルビ除去式の生成を行う.In National Diet Library, books which are possessed in library as "the digital library from meiji era" are open to the public on WEB. Since these are shown as image data and cannot search using document contents, an automatic text conversion is needed. However, ruby is a disturbing text conversion. Since existing techniques of linearly removing ruby had developed for books of the current standard, the techniques are inapplicable to early-modern Japanese books, which have a specific characteristic different from characters of current books. In this paper, we propose a method to remove ruby from early-modern Japanese books using Genetic Programming., Aug. 2013, 6, 2, 53-62, 62
  • Refereed, 情報処理学会論文誌 数理モデル化と応用, The Development and Evaluation of a Sumiji Learning Support System for Visually Impaired People, TAKATA Masami, Aug. 2013, 6, 2, 128-136
  • Refereed, In Proceedings of 2013 International Conference on Parallel and Distributed Processing Techniques and Applications, Locality Analysis for Characterizing Applications Based on Sparse Matrices, TAKATA Masami; Noboru Tanabe; Sonoko Tomimori; Masami Takata; Kazuki Joe, Jul. 2013, II, 299-315
  • Refereed, In Proceedings of 2013 International Conference on Parallel and Distributed Processing Techniques and Applications, Information Processing Society of Japan (IPSJ), Improved Computation of Bounds for Positive Roots of Polynomials, TAKATA Masami; Masami Takata; Takuto Akiyama; Sho Araki; Kinji Kimura; Yoshimasa Nakamur, A new lower bound for computing positive roots of polynomial equations is proposed. We discuss a two-stage algorithm for computing positive roots of polynomial equations. We employ the new bound to accelerate the continued fraction method based on Vincent's theorem. Finally, we conduct experiments to evaluate the effectiveness of the proposed lower bound., Jul. 2013, II, 8, 168-174, 4
  • Refereed, 情報処理学会SACCISシンポジウム, Xeon PhiのGather命令とHMCのGatherの差異, TAKATA Masami, May 2013, 2013, 126-127, 127
  • Refereed, 情報処理学会SACCISシンポジウム, 疎行列キャッシュに基づくGraph500ベンチマークの特性解析, TAKATA Masami, May 2013, 2013, 158-161
  • Refereed, The Fifth International Conference on future Computational Technologies and Applications, Future Irregular Computing with Memory Accelerators, TAKATA Masami; Noboru Tanabe; Junko Kogou; Sonoko Tomimori; Masami Takata; Kazuki Joe, May 2013, 74-80
  • Refereed, 情報処理学会論文誌 数理モデル化と応用, 情報処理学会, 古典油彩画グレーズ技法シミュレーションの高速化, TAKATA Masami, コンピュータ上での絵具の発色シミュレーションは,グラフィックソフトウェアへの応用が期待できる.古典油彩画における「グレーズ」は,薄く溶いた透明な絵具の層を重ねる技法であり,混色で表現できない色調を,重色によって表現するものである.その発色は,放射伝達方程式を解くことによって求められる.しかし従来手法は,異なる絵具でグレーズを行う方法が示されておらず,高精度な解を得るには計算量が大きくなるため,グラフィックソフトウェアへの応用には不向きである.本研究では,まず従来手法を異なる絵具の層に対応できるように拡張する.さらに,係数行列を分解し,刻み幅を固定することにより,高精度な解を小さい計算量で得ることが可能な数値解法を提案する.また,実際の油絵具のデータを用いて数値実験を行い,高精度な解が,従来手法の約4%~7%の計算時間で求められることを示す.Computer simulations of paint color formation are potentially applicable to graphics software. In the "glazing" technique of the classical oil painting, one or more thin transparent layers of paints are applied to an opaque layer to create delicate color tones that cannot be achieved by color mixture. Its coloring can be simulated by solving a radiative transfer equation. However, the conventional technique is not suitable for application to graphics software because it is limited to one color and computational complexity rapidly increases with precision. Accordingly, we have developed a technique for multiple different paints, and also propose a numeric solution with high precision and small computational complexity by decomposing a coefficient matrix and fixing a step size. Furthermore, numerical simulations using data of actual oil paints shows that computation time for highly precise solutions is reduced to about 4%~7% compared to the conventional technique., Mar. 2013, 6, 1, 119-127, 127
  • Refereed, JSIAM Letters, The Japan Society for Industrial and Applied Mathematics, An application of the Kato-Temple inequality on matrix eigenvalues to the dqds algorithm for singular values, TAKATA Masami; Takumi Yamashita; Kinji Kimura; Masami Takata; Yoshimasa Nakamura, Choice of suitable shifts strongly influences performance of numerical algorithms with shift for computing matrix eigenvalues or singular values. On the dqds (differential quotient difference with shifts) algorithm for singular values, a new shift strategy is proposed in this paper. The new shift strategy includes shifts obtained from an application of the Kato-Temple inequality on matrix eigenvalues. The dqds algorithm with the new shift strategy is shown to have a better performance in iteration number than that of the subroutine DLASQ in LAPACK., Mar. 2013, 5, 21-24, 24
  • Refereed, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Character of graph analysis workloads and recommended solutions on future parallel systems, Noboru Tanabe; Sonoko Tomimori; Masami Takata; Kazuki Joe, Graph500 is a benchmark suite for big data analysis. Matrices used for Graph500 inherit the properties of graph analysis such as breadth first search for SNS and PageRank for web searching engine. Especially power saving is very important for its execution on future massively parallel processors and clouds. The spatial locality of sparse matrices used for Graph500 and its behaviors on cache memory are investigated. The experimental results show the spatial locality of sparse matrices used for Graph500 is very low. It is very difficult to solve the problem by just software approach because of the huge size and the randomness of their accesses. Therefore, we recommend hardwired scatter/gather functions at memory side. They improve the processing speed in an order of magnitude. For achieving both of low power and high throughput of random access, we recommend implementing hardwired scatter/gather functions on logic-base in Hybrid Memory Cube (HMC). We also describe brief considerations of the power saving in the case of low cache hit rate application such as graph500. For example, when the hit rate is 15%, the power saving ratio of memory access is about 30-fold. © Springer International Publishing Switzerland 2013., 2013, 8285, 1, 402, 415, International conference proceedings
  • Refereed, 情報処理学会HPCシンポジウム, Graph500競争を独走するための未開拓領域の予備評価, TAKATA Masami, Jan. 2013, 2013, 85-85, 85
  • Refereed, 情報処理学会HPCシンポジウム, 特異値計算アルゴリズムdqds法およびmdLVs法のための新しいシフト戦略, TAKATA Masami, Jan. 2013, 2013, 1-10
  • Refereed, HIGH PERFORMANCE COMPUTING FOR COMPUTATIONAL SCIENCE - VECPAR 2012, SPRINGER-VERLAG BERLIN, Accelerating the Reorthogonalization of Singular Vectors with a Multi-core Processor, Hiroki Toyokawa; Hiroyuki Ishigami; Kinji Kimura; Masami Takata; Yoshimasa Nakamura, The dLV twisted factorization is an algorithm to compute singular vectors for given singular values fast and in parallel. However the orthogonality of the computed singular vectors may be worse if a matrix has clustered singular values. In order to improve the orthogonality, reorthogonalization by, for example, the modified Gram-Schmidt algorithm should be done. The problem is that this process takes a longer time. In this paper an algorithm to accelerate the reorthogonalization of singular vectors with a multi-core processor is devised., 2013, 7851, 379, 390, International conference proceedings
  • Refereed, In Proceedings of JSST 2012 International Conference on Simulation Technology, Real-Time Polygon Reconstruction for Digital Archives of Cultural Properties, TAKATA Masami; Megumi Okumoto; Yuri Iwakata; Asuka Komeda; Tomoko Yonezawa; Masami Takata; Kazuki Joe, Sep. 2012, 429-435
  • Refereed, In Proceedings of JSST 2012 International Conference on Simulation Technology, Manipulation of a VR object using user’s pre-motion, TAKATA Masami; Shiori Mizuno; Asuka Komeda; Tomoko Yonezawa; Naoko Yoshii; Masami Takata; Kazuki Joe, Sep. 2012, 436-442
  • Refereed, 情報処理学会論文誌 コンピューティングシステム, 情報処理学会, 長行を折畳む疎行列ベクトル積方式とGather機能付メモリによる高速化, TAKATA Masami, 本論文では,疎行列ベクトル積のベクトルがデバイスメモリに入りきらないほど大きな問題向けの並列処理方式を提案する.提案手法はGather機能を有する大容量機能メモリ(メモリアクセラレータ)が実現できると仮定し,これが整列したデータをGPUがアクセスする.長い行を適切な折り目で折り畳む提案アルゴリズム(Fold法)が負荷分散を改善し並列性を高める.これが生成した行列を転置して用いる方式はGPU向けのアクセス順序にしている.フロリダ大学の疎行列コレクションを用いて提案方式の性能評価を行った.その結果,間接アクセスの直接アクセス化により,単体性能は既存研究の最大4.1倍に向上した.GPU内キャッシュが溢れる心配もない.GPU間の1対1通信を完全に排除可能にした構成によりスケーラビリティは保証されており,機能メモリとのインタフェースのバースト転送バンド幅で制約される単体性能にノード数を乗じたものが並列実効性能となる.In this paper, we propose a parallel processing strategy for huge scale sparse matrix-vector product whose vector cannot be held on a device memory. The strategy uses a system with GPUs accessing data aligned by functional memories named Memory Accelerator with gather function. This strategy assumes the feasibility of the Memory Accelerator. Proposed algorithm named "Fold method" improves load distribution and parallelism. Transposing matrix produced by it improves access sequence for GPU. We evaluate the performance of proposed strategy with University of Florida Sparse Matrix Collection. The result shows the 4.1 times acceleration over the existing performance record with a GPU in the maximum case. There is no risk of performance degradation by overflowing cache capacity on GPU. Because of the architecture without inter-GPU communications, scalability is guaranteed. Therefore, parallel effective performance is the product of number of nodes and single GPU performance limited by burst transfer bandwidth of interface of functional memory., Aug. 2012, 5, 4, 112-124, 124
  • Refereed, In Proceedings of 2012 International Conference on Parallel and Distributed Processing Techniques and Applications, Performance Evaluation of Some Inverse Iteration Algorithms on PowerXCellTM 8i Processor, TAKATA Masami; Masami Takata; Hiroyuki Ishigami; Kinji Kimura; Yoshimasa Nakamura, Jul. 2012, II, 711-717
  • Refereed, In Proceedings of 2012 International Conference on Parallel and Distributed Processing Techniques and Applications, Real Time Spatiotemporal Biological Stress Level Checking, TAKATA Masami; Marina Uchimura; Yuki Eguchi; Minami Kawasaki; Yoshii Naoko; Tomohiro Umeda; Masami Takata; Kazuki Joe, We are going to investigate and design a healthcare navigation system that consists of sensor devices for human's vital sign, a mobile terminal for transferring the vital data to the cloud, and a cloud computing environment for intelligential processing of the vital data. In this paper, we present a tool for the mobile terminal, i.e. a smartphone, which displays user's biological stress levels. Although the tool is a part of our healthcare navigation system, it does not require the computation resource of the cloud because the stress levels, LF/HF, are enough computable on the processor of the current smartphone in real time. LF/HF is an index of sympathetic nervous activity calculated with the heart rate data obtained from heartbeat sensors. The LF/HF values are displayed in a form of bar graph at user's current location on the Google map of the smartphone. The location information is easily obtained from the GPS of the smartphone. Using the tool, the user can see his/her current stress levels as bar graph on the Google map in real-time. Namely, the tool provides users with their spatiotemporal biological stress levels in real time.We are going to investigate and design a healthcare navigation system that consists of sensor devices for human's vital sign, a mobile terminal for transferring the vital data to the cloud, and a cloud computing environment for intelligential processing of the vital data. In this paper, we present a tool for the mobile terminal, i.e. a smartphone, which displays user's biological stress levels. Although the tool is a part of our healthcare navigation system, it does not require the computation resource of the cloud because the stress levels, LF/HF, are enough computable on the processor of the current smartphone in real time. LF/HF is an index of sympathetic nervous activity calculated with the heart rate data obtained from heartbeat sensors. The LF/HF values are displayed in a form of bar graph at user's current location on the Google map of the smartphone. The location information is easily obtained from the GPS of the smartphone. Using the tool, the user can see his/her current stress levels as bar graph on the Google map in real-time. Namely, the tool provides users with their spatiotemporal biological stress levels in real time., Jul. 2012, II, 12, 744-750, 6
  • Refereed, In Proceedings of 2012 International Conference on Parallel and Distributed Processing Techniques and Applications, AR Based Spatial Reasoning Capacity Training for Students, TAKATA Masami; Mai Hatano; Tomoko Yonezawa; Naoko Yoshii; Masami Takata; Kazuki Joe, In this paper, we propose two methods to train students' spatial reasoning capacity using AR (Augmented Reality). The first method supports students for rotating spatial objects more easily with two AR markers. One marker is used for questions, on which several blocks and a landmark (with a shape of a chick) are displayed. The other marker is used for answers, on which blocks are moved freely. The layout of the blocks toward the chick is selected on the marker. The second method includes limitation of rotation on the marker using some Arduino based hardware. The second method supports students for rotating spatial objects partially. To validate the effect of the trained and resultant spatial reasoning capacity, we perform an experiment using the first method. The analysis results explain the spatial object recognition accuracy increases using the AR learning. To validate the effect of rotation angles, we perform other experiments using the second method. The analysis result shows the rotation angle of sixty degrees is the best for the training of spatial reasoning capacity.In this paper, we propose two methods to train students' spatial reasoning capacity using AR (Augmented Reality). The first method supports students for rotating spatial objects more easily with two AR markers. One marker is used for questions, on which several blocks and a landmark (with a shape of a chick) are displayed. The other marker is used for answers, on which blocks are moved freely. The layout of the blocks toward the chick is selected on the marker. The second method includes limitation of rotation on the marker using some Arduino based hardware. The second method supports students for rotating spatial objects partially. To validate the effect of the trained and resultant spatial reasoning capacity, we perform an experiment using the first method. The analysis results explain the spatial object recognition accuracy increases using the AR learning. To validate the effect of rotation angles, we perform other experiments using the second method. The analysis result shows the rotation angle of sixty degrees is the best for the training of spatial reasoning capacity., Jul. 2012, II, 11, 751-757, 6
  • Refereed, 情報処理学会SACCISシンポジウム, 大規模疎行列やメモリシステムの特性を考慮した高速な汎用疎行列ライブラリの実現に向けて, TAKATA Masami, May 2012, 2012, 65-66, 66
  • Refereed, 情報処理学会HPCシンポジウム, Gather機能を有するメモリアクセラレータの疎行列計算への応用, TAKATA Masami, Jan. 2012, 2012, 5, 32-41
  • Refereed, ASIASIM 2012, PT I, SPRINGER-VERLAG BERLIN, A Multi-resolution Display Method for Digital Archives of Cultural Properties Using Polygon Reconsruction, Megumi Okumoto; Yuri Iwakata; Asuka Komeda; Tomoko Yonezawa; Masami Takata; Kazuki Joe, In this paper, we propose a multi-resolution display method for digital archives of cultural properties using polygon reconstruction. When digital archives of cultural properties are displayed in VR systems for research purpose, mesh resolution had to be changeable on the demand of user. Additionally, for user's comfortability, polygon reconstruction needs to be executed in real-time. History of preliminary polygon reductions is adopted for the proposed method so that polygon reconstruction is executed in real-time. To validate the VR system using the polygon reconstruction based multi-resolution display method, some experiments are performed. As the result of the experiments, the proposed method is performed within 1.0 second, which is considered as "real-time" by the definition of typical user-interface., 2012, 323, Part 1, 190, +, International conference proceedings
  • Refereed, ASIASIM 2012, PT I, SPRINGER-VERLAG BERLIN, Pre-motion Based VR Object Manipulation Definitions and Preliminary Experiments, Shiori Mizuno; Asuka Komeda; Naoko Yoshii; Tomoko Yonezawa; Masami Takata; Kazuki Joe, In this paper, we propose a new method to manipulate objects by using user's pre-motion in a VR environment. Users are provided with an instinctually easy interface to VR with using their natural behaviors as their commands for manipulating VR objects. Hand pre-shaping is known as human's unconscious behavior to face a small object to be grasped. The definitions of pre-motions are determined by users' initial behaviors. Therefore, we present the definitions to classify characteristics of the kinds of objects and pre-motions. Using the above definitions, we develop a prototype system to validate the classifications. Consequently, manipulating VR objects by pre-motion is possible and promising., 2012, 323, Part 1, 499, +, International conference proceedings
  • Refereed, 5TH BIOMEDICAL ENGINEERING INTERNATIONAL CONFERENCE (BMEICON 2012), IEEE, Constitution and Phase Analysis of Alpha Waves, Yu Ishikawa; Masami Takata; Kazuki Joe, Alpha waves are electroencephalogram discovered by Hans Berger in 1929 and have been studied by many researchers. Recently the amplitude and the phase of alpha waves attract attention, and various studies are reported. So far the alpha waves are not concretely modeled to explain how they are constituted. In this paper, we analyze the phase of alpha waves for their future modeling. As pre-processing, we extract alpha wave data from EEGs and divide the data into alpha wave constituent waves. Analyzing the feature quantity of the alpha wave constituent waves, it turns out that the amplitude and the wavelength of alpha wave constituent waves are correlated with a constant time delay., 2012, 1-5, International conference proceedings
  • Refereed, NVIDIA GTC Workshop Japan 2011, GPGPU向けメモリアクセラレータ, 田邊昇; Boonyasitpichai Nuttapon; 中條拓伯; 小郷絢子; 高田雅美; 城和貴, Jul. 2011, Scientific journal
  • Refereed, In Proceedings of 2011 International Conference on Parallel and Distributed Processing Techniques and Applications, An Improved Shift Strategy for the Modified Discrete Lotka-Volterra with Shift Algorithm, TAKATA Masami; Masami Takata; Takumi Yamashita; Akira Ajisaka; Kinji Kimura; Yoshimasa Nakamura, We propose a new mathematical shift strategy for the modified discrete Lotka-Volterra with shift (mdLVs) algorithm. The mdLVs algorithm computes the singular values of bidiagonal matrices. It is known that the convergence of the mdLVs algorithm is accelerated when the shift is close to and less than the square of the smallest singular value of the input matrix. In the original mdLVs algorithm, the Johnson bound is adopted. Our improved mdLVs algorithm combines the Gerschgorin-type bound, the Kato-Temple bound, the Laguerre shift, and the generalized Newton shift. For different combinations, we discuss the computational time and number of iterations.We propose a new mathematical shift strategy for the modified discrete Lotka-Volterra with shift (mdLVs) algorithm. The mdLVs algorithm computes the singular values of bidiagonal matrices. It is known that the convergence of the mdLVs algorithm is accelerated when the shift is close to and less than the square of the smallest singular value of the input matrix. In the original mdLVs algorithm, the Johnson bound is adopted. Our improved mdLVs algorithm combines the Gerschgorin-type bound, the Kato-Temple bound, the Laguerre shift, and the generalized Newton shift. For different combinations, we discuss the computational time and number of iterations., Jul. 2011, II, 4, 720-726, 6
  • Refereed, In Proceedings of 2011 International Conference on Parallel and Distributed Processing Techniques and Applications, Evaluation of the SVM based Multi-Fonts Kanji Character Recognition Method for Early-Modern Japanese Printed Books, TAKATA Masami; Manami Fukuo; Yurie Enomoto; Nakao Yoshii; Masami Takata; Tsukasa Kimesawa; Kazuki Joe, Jul. 2011, II, 727-732
  • Refereed, In Proceedings of 2011 International Conference on Parallel and Distributed Processing Techniques and Applications, Optimization of the Particle-based Volume Rendering for GPUs with Hiding Data Transfer Latency, TAKATA Masami; Kyoko Nakao; Erika Matsui; Naoko Yoshii; Masami Takata; Kazuki Joe, Jul. 2011, II, 733-738
  • Refereed, In Proceedings of 2011 International Conference on Parallel and Distributed Processing Techniques and Applications, A Real-time Analysis Environment for a Wireless BMI Device Enobio, TAKATA Masami; Yu Ishikawa; Sanae Teramae; Naoko Yoshii; Masami Takata; Kazuki Joe, Jul. 2011, II, 739-744
  • Refereed, IPSJ Online Transactions, Information Processing Society of Japan, On Auto-tuned Pre/postprocessing for the Singular Value Decomposition of Dense Square Matrices, TAKATA Masami, An auto-tuning technique is devised for fast pre/postprocessing for the singular value decomposition of dense square matrices with the Dongarra or the Bischof-Murata algorithms. The computation speed of these two algorithms varies depending on a parameter and specification of computers. By dividing these algorithms into several parts and by modeling each of them, we can estimate their computation times accurately. This enables us to choose an optimal parameter and the faster algorithm prior to execution. Consequently the pre/postprocessing is done faster and the singular value decomposition is applied faster to dense square matrices. Numerical experiments show the effectiveness of the proposed auto-tuning function. The I-SVD library, which incorporates this auto-tuning function, has been published., May 2011, 4, 3, 9-21, 146
  • Refereed, PROCEEDINGS OF THE FIRST WORKSHOP ON IRREGULAR APPLICATIONS: ARCHITECTURES AND ALGORITHM (IAAA'11), ASSOC COMPUTING MACHINERY, A Memory Accelerator with Gather Functions for Bandwidth-bound Irregular Applications, Noboru Tanabe; Boonyasitpichai Nuttapon; Hironori Nakajo; Yuka Ogawa; Junko Kogou; Masami Takata; Kazuki Joe, Compute intensive processing can be easily accelerated using processors with many cores such as GPUs. However, memory bandwidth limitation becomes serious year by year for memory bandwidth intensive applications such as sparse matrix vector multiplications (SpMV). In order to accelerate memory bandwidth intensive applications, we have proposed a memory system with additional functions of scattering and gathering. For the preliminary evaluation of our proposed system, we assumed that the throughput of the memory system was sufficient. In this paper, we propose a memory system with scattering and gathering using many narrow memory channels. We evaluate the feasible throughput of the proposed memory system based on DDR3 DRAM with the modified DRAMsim2 simulator. In addition, we evaluate the performance of SpMV using our method for the proposed memory system and a GPU. We have confirmed the proposed memory system has good performance and good stability for matrix shape variation using fewer pins for external memory., 2011, 35, 42, International conference proceedings
  • Refereed, CHEMICAL PHYSICS LETTERS, ELSEVIER SCIENCE BV, Analysis of motion features for molecular dynamics simulation of proteins, Mayumi Kamada; Mikito Toda; Masakazu Sekijima; Masami Takata; Kazuki Joe, Recently, a new method for time series analysis using the wavelet transformation has been proposed by Sakurai et al. We apply it to molecular dynamics simulation of Thermomyces lanuginosa lipase (TLL). Introducing indexes to characterize collective motion of the protein, we have obtained the following two results. First, time evolution of the collective motion involves not only the dynamics within a single potential well but also takes place wandering around multiple conformations. Second, correlation of the collective motion between secondary structures shows that collective motion exists involving multiple secondary structures. We discuss future prospects of our study involving 'disordered proteins'. (C) 2010 Elsevier B.V. All rights reserved., Jan. 2011, 502, 4-6, 241, 247, Scientific journal
  • Refereed, PROCEEDINGS OF THE 19TH INTERNATIONAL EUROMICRO CONFERENCE ON PARALLEL, DISTRIBUTED, AND NETWORK-BASED PROCESSING, IEEE COMPUTER SOC, Scaleable Sparse Matrix-Vector Multiplication with Functional Memory and GPUs, Noboru Tanabe; Yuuka Ogawa; Masami Takata; Kazuki Joe, Sparse matrix-vector multiplication on GPUs faces to a serious problem when the vector length is too large to be stored in GPU's device memory. To solve this problem, we propose a novel software-hardware hybrid method for a heterogeneous system with GPUs and functional memory modules connected by PCI express. The functional memory contains huge capacity of memory and provides scatter/gather operations. We perform some preliminary evaluation for the proposed method with using a sparse matrix benchmark collection. We observe that the proposed method for a GPU with converting indirect references to direct references without exhausting GPU's cache memory achieves 4.1 times speedup compared with conventional methods. The proposed method intrinsically has high scalability of the number of GPUs because intercommunication among GPUs is completely eliminated. Therefore we estimate the performance of our proposed method would be expressed as the single GPU execution performance, which may be suppressed by the burst-transfer bandwidth of PCI express, multiplied with the number of GPUs., 2011, 101, 108, International conference proceedings
  • Refereed, In Proceedings of 2010 International Conference on Parallel and Distributed Processing Techniques and Applications, 情報処理学会, Multi-Regression Analysis of Music Impressions for Music Evaluation, TAKATA Masami; Tomomi Yamanaka; Kyoko Komatsu; Naoko Yoshii; Masami Takata; Kazuki Joe, 本研究では,ユーザが求める曲をコンピュータが探索,提示するための手法を提案する.ユーザにより入力された音楽の印象をもとに,コンピュータが曲の特徴を計算する.コンピュータを用いて音楽を評価するために,SD 法による音楽評価実験を行う.そして重回帰分析を用いてその関係をモデル式として表す.このモデル式は音楽の印象から音楽の特徴を計算する.特徴の計算にはマルチエージェントを使用する.算出したモデル式を各エージェントに実装させ,エージェント間の協調により特徴の計算を行う.本手法を用いることによりコンピュータを用いて音楽の検索を行うことが可能となる.In this paper, we propose a music evaluation method. In this method, impressions of given music are represented by features by value, so that the music impressions are processed on computers. To obtain the music impressions, we experiment music impressions by using a semantic differential method. Then, we construct several models based on the experimental results with multiregression analysis. By using these models, music features are uniquely digitalized. A multi-agent system is expected to use the music feature. Each agent of the multi-agent adopts a different model to cooperate between agents. Then, users can understand their music preference based on the music features values., Jul. 2010, II, 12, 693-702, 6
  • Refereed, In Proceedings of 2010 International Conference on Parallel and Distributed Processing Techniques and Applications, A Music Composition Model with Genetic Programming, TAKATA Masami; Kyoko Komatsu; Tomomi Yamanaka; Masami Takata; Kazuki Joe, Jul. 2010, II, 686-692
  • Refereed, In Proceedings of 2010 International Conference on Parallel and Distributed Processing Techniques and Applications, Optimization of GROMACS for CUDA Environments, TAKATA Masami; Noriko Hata; Masami Takata; Kazuki Joe, Jul. 2010, II, 679-685
  • Refereed, In Proceedings of 2010 International Conference on Parallel and Distributed Processing Techniques and Applications, Update Dependence Analysis of Web Pages for Incremental Crawling, TAKATA Masami; Naomi Hiromichi; Masami Takata; Kazuki Joe, Jul. 2010, II, 672-678
  • Refereed, In Proceedings of 2010 International Conference on Parallel and Distributed Processing Techniques and Applications, 情報処理学会, Feature Detection of Volume Data based on Topology and its Application to Database Construction, TAKATA Masami; Keiko Kuriu; Masami Takata; Kazuki Joe, 本研究では,ボリュームデータから位相幾何学的に特徴のある部分を抽出し,それらを集めデータベースを設計する.あらかじめ位相幾何学な特徴を意義し,解析するデータ内で,定義した特徴部の有無を確認する.そして,特徴部の有無を解析したデータのメタデータとして扱い,データベースを設計する.それによって,特徴部をクエリとして検索にかけ,対応データやパラメータの値を返すことで,解析者にとって取得したい興味分野の検索補助が可能となる.In this paper, propose a topological feature extraction method for volume data, and develop a database, which collects that kind of extracted feature. First, we define topological feature. Next, we extract topological features of a volume data set. In the case that topological features are extracted, the features are given as metadata of the volume data. A database is constructed with the volume data and the metadata. By using the database, when features are obtained as queries, the related data and parameters are returned. Hence, the database can find out the interest features with topological visualization., Jul. 2010, II, 6, 665-671, 6
  • Refereed, In Proceedings of 2010 International Conference on Parallel and Distributed Processing Techniques and Applications, Speed-up in mdLVs by Limitation in Computational Number of Shift, TAKATA Masami; Masami Takata; Kinji Kimura; Yoshimasa Nakamura, Jul. 2010, II, 619-625
  • Refereed, 情報処理学会論文誌 コンピューティングシステム, 情報処理学会, 密正方行列特異値分解における並列I-SVD法の特性を用いた後処理の高速化, TAKATA Masami, 本稿では,並列 I-SVD 法を用いて密正方行列の特異値分解を行う.そのために,Bischof らのアルゴリズムと村田法を前処理および後処理として用いる.この際,村田法の逆変換に用いる行列の作成と並列 I-SVD 法を同時に行う.これにより,前処理および後処理を含めた特異値分解の計算時間を短縮することが可能となる.We compute the singular value decomposition for dense square matrices with the parallel I-SVD algorithm. For this purpose, we apply the Bischof's algorithm and Murata's algorithm as pre-processing and post-processing. Furthermore construction of the matrices used for the inverse transformation of Murata's algorithm, and the parallel I-SVD algorithm are performed simultaneously. As a result, the number of idle computation cores decreases, and SVD is achieved faster., Jun. 2010, 3, 2, 30-38, 38
  • Refereed, 情報処理学会SACCISシンポジウム, GPUと機能メモリを用いたヘテロシステムによるスケーラブルな疎行列ベクトル積高速化の提案, TAKATA Masami, May 2010, 2010, 5, 109-110, 110
  • Refereed, 情報処理学会論文誌 数理モデルと応用, 情報処理学会, 電車ノイズを含む地電位差データからの矩形状地震前駆的シグナル自動抽出, TAKATA Masami, 地電位差データに稀に含まれる異常電流を地震前駆的シグナル (SES: Seismic Electric Signals) として検出することで地震の予知を図る手法がある.強い電車ノイズを含む地電位差データから SES を検出するために,独立成分分析 (ICA: Independent Component Analysis) を地電位差データに適用し,電車ノイズを分離することが試みられており,結果として ICA が有効であることは確認されている.また,地震予測を迅速に行うためには,ノイズを分離したデータから自動的に SES を検出する手法が必要である.本稿では,ICA を用いて電車ノイズを分離した結果から,自動的に SES を検出する手法について提案する.新潟県中越地震の SES を含むと考えられる地電位差データに対して提案手法を適用する実験を行い,適用結果と地震発生データの相関調査を行う.Earthquake predictions are tried by extracting seismic electric signals (SES), which are sometimes contained in telluric current data (TCD). TCD observed in Japan is also containing strong train noise. To separate train noise from TCD, independent component analysis (ICA) has been applied for TCD. The result shows high effect of ICA to separate train noise. However, SES have not been detected in the experiments. In this paper, we approach auto-detection of SES from data, which are separated TCD by ICA. As an experiment of our method, we apply the method to TCD, which may contain SES of The Mid Niigata Prefecture Earthquake in 2004. We also investigate correlation between the result of the experiments and information of earthquake occurrence., Mar. 2010, 3, 2, 22-31, 31
  • Seibutsu Butsuri, The Biophysical Society of Japan General Incorporated Association, 3P046 Time-series Analysis of Protein Structure using Feature of Motion(Protein: Structure & Function,The 48th Annual Meeting of the Biophysical Society of Japan), Kamada Mayumi; Toda Mikito; Sekijima Masakazu; Takata Masami; Joe Kazuki, 2010, 50, 2, S153
  • Refereed, 情報処理学会HPCシンポジウム, 密正方行列特異値分解における並列I-SVD法の特性を用いた後処理の高速化, TAKATA Masami, Jan. 2010, 2010, 43-50
  • Refereed, In Proceedings of 2009 International Conference on Parallel and Distributed Processing Techniques and Applications, Clustering the Temporal Sequences Feature of 3D Protain Structure, TAKATA Masami; Mayumi Kamada; Sachi Kimura; Mikito Toda; Masami Takata; Kazuki Joe, Jul. 2009, II, 749-755
  • Refereed, In Proceedings of 2009 International Conference on Parallel and Distributed Processing Techniques and Applications, Optimization of a GP Application for the Cell Processor, TAKATA Masami; Yuri Shirakawa; Chiho Misaka; Chisato Ishikawa; Masami Takata; Kazuki Joe, Jul. 2009, II, 742-748
  • Refereed, In Proceedings of 2009 International Conference on Parallel and Distributed Processing Techniques and Applications, Modeling the Developing Process of Depth Perception for Elementary Students, TAKATA Masami; Naoko Yoshii; Masami Takata; Kazuki, Jul. 2009, II, 735-741
  • Refereed, In Proceedings of 2009 International Conference on Parallel and Distributed Processing Techniques and Applications, Recognition of Multi-Fonts Character in Early-Modern Printed Books, TAKATA Masami; Chisato Ishikawa; Naomi Ashida; Yurie Enomoto; Masami Takata; Tsukasa Kimesawa; Kazuki Joe, Jul. 2009, II, 728-734
  • Refereed, In Proceedings of 2009 International Conference on Parallel and Distributed Processing Techniques and Applications, Deterioration Lawns Model Constructed from Image Data, TAKATA Masami; Yurie Enomoto; Chisato Ishikawa; Masami Takata; Kazuki Joe, Jul. 2009, II, 721-727
  • Refereed, In Proceedings of 2009 International Conference on Parallel and Distributed Processing Techniques and Applications, 情報処理学会, On Parallelization of the I-SVD Algorithm and its Evaluation for Clustered Singular Values, TAKATA Masami; Hiroki Toyokawa; Kinji Kimura; Masami Takata; Yoshimasa Nakamura, 行列の特異値分解を行うアルゴリズムの 1 つとして I-SDD 法がある.これは,特異値計算に mdLVs 法,特異ベクトル計算に dLV 型ツイスト分解を用いたアルゴリズムであり,計算速度や相対精度などにおいて QR 法に基づく標準解法より優れている.最近の研究において,I-SVD 法は内部で用いられている mdLVs 法と dLV 型ツイスト分解の処理を適切に複数の計算コアに割り当てることによって,マルチコアプロセッサ上で効率的な並列化を行うことができた.しかし,近接特異値を持つ行列に対して dLV 型ツイスト分解を行うと,特異ベクトルの直交性が悪くなることが知られている.本研究では,あまり大きくない大きさのクラスターを複数持つ行列について特異ベクトルの直交性を改善することによって,並列化された I-SVD 法の改良を試みる.また数値実験を行い,改良された並列版 I-SVD 法を評価する.A parallelization of the I-SVD algorithm is performed. Here, the I-SVD algorithm is a new singular value decomposition (SVD) algorithm consisting of the mdLVs scheme and the dLV twisted factorization. For some random matrices, this algorithm is better than today's standard algorithms for SVD, with some respects. However, if some singular values are not separated well and form clusters, the orthogonality of the singular vectors computed by the twisted factorization may be worse. In this paper, the parallelization of the I-SVD is modified to improve the orthogonality of such singular vectors in terms of a reorthogozalization process. In order to evaluate this modified algorithm, numerical experiments are conducted with a multi-core processor. As a result, the I-SVD algorithm is parallelized for matrices whose singular values form clusters. It is shown that a more efficiency of the parallelization results for matrices having many clusters of medium or small size., Jul. 2009, II, 10, 711-717, 6
  • Refereed, In Proceedings of 2009 International Conference on Parallel and Distributed Processing Techniques and Applications, Speed-up in mdLVs by Limitation in Computational Number of Shift, TAKATA Masami; Masami Takata; Kinji Kimura; Yoshimasa Nakamura, Jul. 2009, II, 704-710
  • Refereed, In Proceedings of 2009 International Conference on Parallel and Distributed Processing Techniques and Applications, The Architecture of Visualization System using Memory with Memory-side Gathering and CPUs with DMA-type Memory Accessing, TAKATA Masami; Noboru Tanabe; Manami Sasaki; Hironori Nakajo; Masami Takata; Kazuki Joe, Jul. 2009, II, 427-433
  • Refereed, 情報処理学会論文誌 数理モデルと応用, iアプリによる数値計算実アプリケーションの実装例, TAKATA Masami, 携帯電話に代表される携帯情報端末は,待機状態における余剰計算リソースを有しており,これらを集約することができれば新たな計算リソースの創出につながる.携帯情報端末のハードウェア技術革新は日進月歩であり,数年前の PC のスペックに迫るような携帯情報端末が出現することは必然である.本論文では,これら携帯情報端末を分散処理システムとして構築していくための一事例として,携帯電話に対する実アプリケーション・プログラムの移植方法と移植した結果の実行性能について報告する.その結果,現状の携帯電話の言語処理系には問題があるものの,近い将来携帯情報端末による分散処理システムの構築が実現可能であることを示す.PDAs such as cellar phones have surplus computing resource when they are in a standby mode. If such surplus computing resource are to be consolidated, we will have a new class of computing resource. The hardware technologies for PDA are innovated day by day, so a highly specialized PDA that is almost equivalent to existing PCs will appear in the near future. In this paper, we propose a distributed processing system model by many PDAs. For the sake of a case study, we explain the porting methods of a real application program to a cellra phone and show some performance results of the ported application. As a result, we find the current language specification for the cellar phone has a restriction, but we are sure that the restriction will disappear in a very near future so that the distributed processing systems by many PDAs are realizable., Mar. 2009, 2, 2, 38-46, 46
  • Refereed, JSIAM Letters, The Japan Society for Industrial and Applied Mathematics, On parallelism of the I-SVD algorithm with a multi-core processor, TAKATA Masami; Hiroki Toyokawa; Kinji Kimura; Masami Takata; Yoshimasa Nakamura, The I-SVD algorithm is a singular value decomposition algorithm consisting of the mdLVs scheme and the dLV twisted factorization. By assigning each piece of computations to each core of a multi-core processor, the I-SVD algorithm is parallelized partly. The basic idea is a use of splitting and deflation in the mdLVs. The splitting divides a bidiagonal matrix into two smaller matrices. The deflation gives one of the singular values, and then the corresponding singular vector becomes computable by the dLV. Numerical experiments are done on a multi-core processor, and the algorithm can be about 5 times faster with 8 cores., Jan. 2009, 1, 48-51, 51
  • Refereed, Transactions of the Japan Society for Industrial and Applied Mathematics, The Japan Society for Industrial and Applied Mathematics, 固有値分解を目的としたツイスト分解法による分割統治法の改善, TAKATA Masami, An algorithm which consists of a simplified D & C and twisted factorization is proposed for symmetric tridiagonal eigenvalue decomposition. The complexity is O(n^2) and the memory usage is O(n) if no cluster exists. The orthogonality can be improved by additional one step of the inverse iteration, although the classical D & C shows better one. In some numerical tests, our algorithm shows stable speed and better accuracy of the decompositions. But the orthogonality is worse than those of the classical D & C in up to three digits., Dec. 2008, 18, 4, 611-630, 630
  • Refereed, Transactions of the Japan Society for Industrial and Applied Mathematics, The Japan Society for Industrial and Applied Mathematics, 特異値分解法I-SVDにおける左特異ベクトル計算部の改善, TAKATA Masami, In this paper, computation of left singular vectors by the I-SVD algorithm for fast bidagonal SVD is improved so that the computed left singular vectors have better orthogonality and accuracy. The efficiency of improved I-SVD is evaluated through numerical experimentations., Sep. 2008, 18, 3, 389-407, 407
  • Refereed, In Proceedings of International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'08), Real-time 3D Movie Generation by Anaglyph for Live Streaming, TAKATA Masami; Yuki Kohor, Jul. 2008, II, 758-763, 763
  • Refereed, In Proceedings of International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'08), Information Processing Society of Japan (IPSJ), Construction of an Ethanol-Tolerant Yeast Model using the E-cell System, TAKATA Masami; Mayumi Kamada他, In brewing Japanese sake, since ethanol reaches to a high concentration by using a peculiar manufacturing method, yeast with strong alcohol tolerance is needed. Such yeast can be obtained through huge number of trial and error processes with huge cost and long time. To avoid this problem, we try to model the response mechanism of alcohol-tolerant yeast by computer simulation based on known parameters from existing literatures. We hope new insight will be obtained by our model. In this paper, we model a simple metabolism of 1,3-β glucan synthesis, which is the major components of cell well, and a signaling pathway of the Pkcl-MAPK cascade using the E-Cell system., Jul. 2008, II, 65, 776-782, 30
  • Refereed, GRAPP 2008: PROCEEDINGS OF THE THIRD INTERNATIONAL CONFERENCE ON COMPUTER GRAPHICS THEORY AND APPLICATIONS, INSTICC-INST SYST TECHNOLOGIES INFORMATION CONTROL & COMMUNICATION, An interactive 3D visualization model by live streaming for remote scientific visualization, Eriko Touma; Satomi Hara; Mari Kurumi; Yuri Shirakawa; Chisato Ishikawa; Masami Takata; Takeshi Horinouchi; Kazuki Joe, Recent improvement of high-end GPUs has made it possible to perform real-time 3D visualization such as volume rendering and 3D contour plot for scientific data locally. A web browser based remote 3D visualization by visualization servers is attractive, but data transfer overhead prevents from performing interactive operations. We propose an interactive remote 3D visualization model by live streaming for geophysical fluids research. In this model, we use live streaming flash media for the web browser based operations keeping minimum quality of data analysis and minimum bit rate for live streaming of flash media. Preliminary experiments with a prototype system validate the effectiveness of our proposing model., 2008, II, 409, 414, International conference proceedings
  • Refereed, 情報処理学会HPCシンポジウム, 悪条件2重対角行列のための特異値分解I-SVD法の改良, TAKATA Masami, Jan. 2008, 2008, 2, 9-16
  • Refereed, 情報処理学会論文誌 数理モデル化と応用, 情報爆発時代に向けた大気科学分野のための3次元可視化モデルの提案, TAKATA Masami, Dec. 2007, 48, SIG19(TOM19), 86-98
  • Refereed, 情報処理学会論文誌 数理モデル化と応用, Information Processing Society of Japan (IPSJ), アブストラクトを用いた原子分子物理学分野の論文分類支援システムの設計と実装, TAKATA Masami, The cross section data of ionization and excitation by collision among atoms, mulecules and electrons is very useful in various research fields. We can obtain such data out of online jounals of atomic and molecular physics research as digital papers. To obtain the data, we need to classify online papers. However, the classification of the online papers is difficult because there are just abstracts available for free on web pages in general. In this paper, we design a paper classification supporting system to find papers including the data, using just abstracts. We apply a machine learning technique, which is a conventional text classification method, to the system in order to prove that our system using just abstracts is efficient., Oct. 2007, 48, SIG15(TOM18), 189-198, 198
  • Refereed, In Proceedings of International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'07), Algorithms for Generating Bidiagonal Test Matrices, TAKATA Masami; Masami Takata, Jun. 2007, II, 732-738, 738
  • Refereed, In Proceedings of International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'07), CSREA Press, Development of Molecular Dynamics Simulation based Flexible Docking System, TAKATA Masami; Sachi Kimura, Jun. 2007, II, 739-745, 745, International conference proceedings
  • Refereed, In Proceedings of International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'07), Construction of Paper Classification System Using SVM, TAKATA Masami; Naomi Ashida他, Jun. 2007, II, 759-765
  • Refereed, In Proceedings of International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'07), 3D Visualization System Gateau for Atmospheric Science: Design Concept and Practical Evaluation, TAKATA Masami; Eriko Toum, Jun. 2007, II, 766-772
  • Refereed, In Proceedings of International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'07), Optimization of Molecular Dynamics Simulation on Cell Processor, TAKATA Masami; Manami Sasaki他, Jun. 2007, II, 780-786
  • Refereed, In Proceedings of International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'07), CSREA Press, Development of an SVM based Prediction System for Metalbinding Sites in Protein, TAKATA Masami; Masami Nakazawa, Jun. 2007, II, 972-978, 978, International conference proceedings
  • Refereed, In Proceedings of International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'07), Auto Classification of Feces for Health Condition Analysis, TAKATA Masami; Miho Komatsubara, Jun. 2007, II, 979-985
  • Refereed, 情報処理学会SACSISシンポジウム, ダブル分割統治法による対称3重対角行列の固有値分解, TAKATA Masami, May 2007, 2007, 5, 121-128
  • Refereed, Proceedings of the IADIS International Conference, Aopplied Computing 2007 (AC2007), DEVELOPMENT OF ORTHOGONALITY OF SINGULAR VECTORS COMPUTED BY I-SVD ALGORITHM, TAKATA Masami; Masami Takata, Feb. 2007, 437-442
  • Refereed, ADVANCES IN MULTIMEDIA MODELING, PT 1, SPRINGER-VERLAG BERLIN, Similarity searching techniques in content-based audio retrieval via hashing, Yi Yu; Masami Takata; Kazuki Joe, With this work we study suitable indexing techniques to support efficient, content-based music retrieval in large acoustic databases. To obtain the index-based retrieval mechanism applicable to audio content, we pay the most attention to the design of Locality Sensitive Hashing (LSH) and the partial sequence comparison, and propose a fast and efficient audio retrieval framework of query-by-content. On the basis of this indexable framework, four different retrieval schemes, LSH-Dynamic Programming (DP), LSH-Sparse DP (SDP), Exact Euclidian LSH ((ELSH)-L-2)-DP, (ELSH)-L-2-SDP, are presented and estimated in order to achieve an extensive understanding of retrieval algorithms performance. The experiment results indicate that compared to other three schemes, (ELSH)-L-2-SDP exhibits best tradeoff in terms of the response time, retrieval ratio, and computation cost., 2007, 4351, Part I, 397, 407, International conference proceedings
  • Refereed, PROCEEDINGS OF THE IASTED INTERNATIONAL CONFERENCE ON PARALLEL AND DISTRIBUTED COMPUTING AND NETWORKS, ACTA PRESS ANAHEIM, Parallelism of double Divide and Conquer algorithm for singular value decomposition, Taro Konda; Hiroaki Tsuboi; Masami Takata; Masashi IIwasaki; Yoshimasa Nakamura, This paper presents some numerical evaluations of parallel double Divide and Conquer for singular value decomposition. For eigenvalue decomposition and singular value decomposition, double Divide and Conquer was recently proposed. It rst computes eigen/singular values by a compact version of Divide and Conquer. The corresponding eigen/singular vectors are then computed by twisted factorization. The speed and accuracy of double Divide and Conquer are as good or even better than standard algorithms such as QR and the original Divide and Conquer. In addition, it is expected that double Divide and Conquer has great parallelism because each step is theoretically parallel and heavy communication is not required. This paper numerically evaluates a parallel implementation of dDC with MPI on some large scale problems using a distributed memory architecture and a massively parallel super computer, especially in terms of parallelism. It shows high scalability and super linear speed-up is observed in some cases., 2007, 393, +, International conference proceedings
  • Refereed, PARALLEL COMPUTING TECHNOLOGIES, PROCEEDINGS, SPRINGER-VERLAG BERLIN, Accelerating the singular value decomposition of rectangular matrices with the CSX600 and the integrable SVD, Yusaku Yamamoto; Takeshi Fukaya; Takashi Uneyama; Masami Takata; Kinji Kimura; Masashi Iwasaki; Yoshimasa Nakamura, We propose an approach to speed up the singular value decomposition (SVD) of very large rectangular matrices using the CSX600 floating point coprocessor. The CSX600-based acceleration board we use offers 50GFLOPS of sustained performance, which is many times greater than that provided by standard microprocessors. However, this performance can be achieved only when a vendor-supplied matrix-matrix multiplication routine is used and the matrix size is sufficiently large. In this paper, we optimize two of the major components of rectangular SVD, namely, QR decomposition of the input matrix and back-transformation of the left singular vectors by matrix Q, so that large-size matrix multiplications can be used efficiently. In addition, we use the Integrable SVD algorithm to compute the SVD of an intermediate bidiagonal matrix. This helps to further speed up the computation and reduce the memory requirements. As a result, we achieved up to 3.5 times speedup over the Intel Math Kernel Library running on an 3.2GHz Xeon processor when computing the SVD of a 100,000 x 4000 matrix., 2007, 4671, 340, +, International conference proceedings
  • Refereed, Proceedings of the First International Workshop on Learning the Semantics of Audio Signals (LSAS 2006), Index-based Similarity Searching with Partial Sequence Comparison for Query-by-Content Audio Retrieval, TAKATA Masami; Yi Yu, Dec. 2006, 76-86, 86
  • Refereed, 2nd IEEE International Conference on e-Science and Grid Computing, Automatic Viewpoint Selection for a Visualization I/F in a PSE, TAKATA Masami; Machiko Nakagawa, Dec. 2006, CD
  • Refereed, IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS, IEICE-INST ELECTRONICS INFORMATION COMMUNICATIONS ENG, Naive mean field approximation for sourlas error correcting code, Masami Takata; Hayaru Shouno; Masato Okada, Solving the error correcting code is an important goal with regard to communication theory. To reveal the error correcting code characteristics, several researchers have applied a statistical-mechanical approach to this problem. In our research, we have treated the error correcting code as a Bayes inference framework. Carrying out the inference in practice, we have applied the NMF (naive mean field) approximation to the MPM (maximizer of the posterior marginals) inference, which is a kind of Bayes inference. In the field of artificial neural networks, this approximation is used to reduce computational cost through the substitution of stochastic binary units with the deterministic continuous value units. However, few reports have quantitatively described the performance of this approximation. Therefore, we have analyzed the approximation performance from a theoretical viewpoint, and have compared our results with the computer simulation., Aug. 2006, E89D, 8, 2439, 2447, Scientific journal
  • Refereed, In Proceedings of International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'06), Information Processing Society of Japan (IPSJ), Verification of dLVv Transformation for Singular Vector Computation with High Accuracy, TAKATA Masami; Masami Takata, In this paper, we first examine a relationship between errors of singular value and the accuracy of singular vector, whici is computed through the twisted factorization by the discrete Lotka- Volterra with variable step-size transformation. Secondly, we discuss a suitable parameter choice by evaluating the orthogonality of singular vectors and the errors of singular value decomposition., Jun. 2006, II, 68, 881-887, 46
  • Refereed, In Proceedings of International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'06), Information Processing Society of Japan (IPSJ), Preliminary Result of Parallel Double Divide and Conquer, TAKATA Masami; Taro Konda, This paper shows a concept for parallelization of double Divide and Conquer and its preliminary result. For singular value decomposition, double Divide and Conquer was recently proposed. It first computes singular values by a compact version of Divide and Conquer. The corresponding singular vectors are then computed by twisted factorization. The speed and accuracy of double Divide and Conquer are as well or even better than standard algorithms such as QR and Divide and Conquer. In addition, it is expected that double Divide and Conquer has great parallelism because each step is theoretically parallel and heavy communication is no required. However, any parallel model of double Divide and Conquer has not been studied yet. In this paper, policy of the parallelization is discussed. Then, a parallel implementation with MPI is tested on a distributed memory parallel computer. It successfully shows a high parallelism., Jun. 2006, II, 68, 888-894, 50
  • Refereed, In Proceedings of International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'06), Information Processing Society of Japan (IPSJ), Evaluation of a New Eigen Decomposition Algorithm for Symmetric Tridiagonal Matrices, TAKATA Masami; Hiroaki Tsuboi, This paper focuses on a new extension version of double Divide and Conquer (dDC) algorithm to eigen decomposition. Recently, dDC was proposed for singular value decomposition (SVD) of rectangular matrix. The dDC for SVD consists of two parts. One is Divide and Conquer (D〓C) for singular value and the other is twisted factorization for singular vector. The memory usage of dDC is smaller than that of D〓C. Both theoretical and running time are also shorter than those of D〓C. In this paper, a new dDC for eigen decomposition is proposed. A shift of origin is introduced into our dDC. By some numerical tests, dDC is evaluated with respect to running time and accuracy., Jun. 2006, II, 68, 832-838, 18
  • Refereed, IPSJ, ACS14, 高速特異値分解のためのライブラリ開発, TAKATA Masami; Masami Takata, May 2006, 47, SIG7(ACS14), 91-104
  • Refereed, IPSJ, ACS14, 分割統治法とツイスト分解による新しい特異値分解アルゴリズム, TAKATA Masami, May 2006, 47, SIG7(ACS14), 81-90
  • Refereed, 情報処理学会HPCシンポジウム, 2重対角行列の特異値分解におけるツイスト分解を用いた特異ベクトル計算法の改良, TAKATA Masami; Masami Takata, May 2006, 2007, 1, 103-110
  • Refereed, PROCEEDINGS OF THE IASTED INTERNATIONAL CONFERENCE ON ADVANCES IN COMPUTER SCIENCE AND TECHNOLOGY, ACTA PRESS ANAHEIM, A new singular value decomposition algorithm suited to parallelization and preliminary results, Taro Konda; Masami Takata; Masashi Iwasaki; Yoshimasa Nakamura, This paper proposes a new singular value decomposition algorithm which can be fully parallelized. I-SVD algorithm was recently proposed using the theory of integrable systems. It computes singular value decompositions much faster than standard algorithms such as QR and Divide and Conquer(D&C), and at the same accuracy level. However, we show that the parallelism of I-SVD is limited to singular vector computation only because it shows seriality in the case of singular value computation. Our concern is a fully parallelizable algorithm whose serial version is as fast and accurate as I-SVD. Compared to the original D&C, where most of the running time is consumed for vector updating during singular vector computation, our new algorithm first computes singular values by the "compact" D&C without computing singular vectors. Thus, the compact D&C is wholly faster than the original one. The corresponding singular vectors are then computed by twisted factorization. The algorithm has great parallelism because each step can be executed parallelly. Our new algorithm is numerically tested on some SVD computations. It is as fast as I-SVD and even much faster than the standard algorithms. Also, its accuracy is as good as that of the others. 505-033 in display pdf, 2006, 79, +, International conference proceedings
  • Refereed, 情報処理学会HPCシンポジウム, 分割統治法とツイスト分解による新しい特異値分解アルゴリズム, TAKATA Masami, Jan. 2006, 2006, 1, 105-112
  • Refereed, 情報処理学会HPCシンポジウム, 2重対角行列のための高速特異値分解の開発, TAKATA Masami, Jan. 2006, 2006, 1, 27
  • Refereed, PROCEEDINGS OF THE IASTED INTERNATIONAL CONFERENCE ON PARALLEL AND DISTRIBUTED COMPUTING AND NETWORKS, ACTA PRESS, Performance of a new scheme for bidiagonal singular value decomposition of large scale, M Takata; K Kimura; M Iwasaki; Y Nakamura, To perform singular value decomposition (SVD) of matrices with high accuracy and high-speed, we developed a library by using an integrable system called the discrete Lotka-Volterra (dLV) system. The most well-known routine for the SVD is DBDSQR provided in LAPACK, which is based on the QRs (QR with shift) algorithm. However, DBDSQR is slow and does not perform well for the case where only a few singular vectors are desirable. Recently a new SVD scheme named Integrable-SVD (I-SVD) was developed. The execution time for the I-SVD scheme based on the dLV system and transformation is less. In this paper, we implement and evaluate the I-SVD scheme. To examine its accuracy, we propose a method for making random upper bidiagonal matrices with desired singular values. The corresponding singular vectors are also accurately obtained. From the experimental results, we conclude that the singular vectors computed by the I-SVD scheme have adequate orthogonality, and the scheme also has high speed and accuracy for both the computed singular values and vectors., 2006, 304, +, International conference proceedings
  • Refereed, IPSJ, ACS11, Implementation and its Evaluations of Routine \\for Computing Singular Values with High Accuracy, TAKATA Masami; Masami Takata, 高精度かつ高速に行列を特異値分解するために,我々は,dLV (離散ロトカ・ボルテラ: discrete Lotka--Volterra) 系による新たな特異値分解ライブラリを開発している.この一環として,本論文では,特異値のみを計算する mdLVs (原点シフト付き modified dLV) 法の実装ならびに性能評価を行う.既存ルーチンとしては,線形数値計算ライブラリ LAPACK における DBDSQR と DLASQ がある.DBDSQR は,QRs 法に基づいた特異値分解ルーチンであるが,大規模特異値計算において計算速度,精度がともに十分でない.一方,DLASQ は,dqds 法に基づいた高速高精度な特異値計算ルーチンであるが,収束性が証明されていない.そのため,現在利用されている計算方法は,QRs 法である.これらに比べ,mdLVs 法では,収束性が証明されており,DLASQ より速度では劣るものの,理論的には,同等以上の高精度性を持つ.本論文では,mdLVs 法を適用するために 4 種類の実装手法を提案する.これらの手法の有効性を調べるために, Aug. 2005, 46, SIG12(ACS11), 299-311
  • Refereed, 情報処理学会SACSISシンポジウム, いくつかの CPU の推奨コンパイラにおける特異値計算ルーチンの実行時間の測定, TAKATA Masami, May 2005, 2005, 5, 212-213
  • Refereed, 情報処理学会ハイパフォーマンスコンピューティングと計算科学シンポジウム, ロトカ・ボルテラ系による特異値計算アルゴリズムの実装と性能評価, 高田 雅美; 岩崎 雅史; 木村 欣司; 中村 佳正, Jan. 2005, HPCS2005, 25, 32
  • Refereed, IPSJ, TOM11, The Possibility for Numerical Calculations with i-APPLI, TAKATA Masami; Masami Takata, 携帯電話会社 NTT Docomo の提供する i アプリは,携帯端末において,様々な Java アプリケーションを実行することが可能である.\ni アプリ対応の携帯端末は,計算機としては貧弱ではあるが,ネットワーク機能・計算リソースを有しており,2003 年 11 月の時点では,その市場稼動台数は,$2,000$ 万台を突破している.本研究では,アイドル状態にある携帯端末において,i アプリを用いた数値計算を行うための分散処理システムによって,計算リソースを有効活用する方法を提案する.個々の携帯端末は,貧弱な計算リソースしか持たないため,処理すべき計算問題は,分散化される必要がある.そこで,我々は,この分散処理のための一つのアプローチとして分散遺伝的アルゴリズムを i アプリによって実装し,計算問題解決のためのプロトタイプを作成した., Jan. 2005, 46, SIG2(TOM11), 44-55
  • Refereed, 情報処理学会HPCシンポジウム, ロトカ・ボルテラ系による特異値計算の実装と評価, TAKATA Masami, Jan. 2005, 2005, 25-32
  • Refereed, PDPTA '05: Proceedings of the 2005 International Conference on Parallel and Distributed Processing Techniques and Applications, Vols 1-3, C S R E A PRESS, An evaluation of singular value computation by the discrete Lotka-Volterra system, M Takata; K Kimura; M Iwasaki; Y Nakamura, The mdLVs (modified discrete Lotka-Volterra with shift) scheme for computing singular values is evaluated. Convergence and stability of the mdLVs scheme are guaranteed. It is shown that numerical accuracy of the scheme is equal to or higher than that of the existing routines of LAPACK., 2005, II, 410, 416, International conference proceedings
  • Refereed, PDPTA '04: PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON PARALLEL AND DISTRIBUTED PROCESSING TECHNIQUES AND APPLICATIONS, VOLS 1-3, C S R E A PRESS, A multi-objective genetic algorithm for program partitioning and data distribution using TVRG, M Takata; T Yamaguchi; C Watanabe; Y Nakamura; K Joe, We propose an algorithm that performs data distribution and parallelization simultaneously. The objectives of the simultaneous algorithm are to reduce the length of critical path and the total memory size. Regardless to say, memory usage for each processor must be balanced. To obtain an optimal solution, we first adopted a branch and bound method. Since the branch and bound method often fails in the case of a large task graph, we adopt a multi-objective genetic algorithm, that provides a near optimal solution. For effective simultaneous partitionings, we employ some edge sorting and ordering methods. The effectiveness of our simultaneous partitioning algorithms is shown by experimental results., 2004, I, 345, 351, International conference proceedings
  • Refereed, 情報処理学会MPSシンポジウム, NCC@i-APPLI : i アプリによる数値計算コンポーネントの構想, TAKATA Masami, Oct. 2003, 2003, 14, 45-52
  • Refereed, IPSJ, ACS1, Parallelization and Evaluation of a Direct Numerical Simulation for a Turbulent Flow using Pseudo Spectral Method, TAKATA Masami; Masami Takata, 自由界面における乱流を計算機で解析する場合,直接数値シミュレーションと呼ばれる手法により数値解を求める場合がある.この数値解の精度をあげるためには,しばしば擬似スペクトル法が用いられる.本研究では,擬似スペクトル法を用いた乱流解析プログラムを分散メモリ環境対象の並列プログラムに変換し,その性能評価を行った.擬似スペクトル法は,流体の時間発展方程式を解く際,フーリエ変換を用いて波数空間で計算を行う.この際,通常の並列化で使われる strip mining による実空間の均等分割によって配列データの分割を行った場合,通信時間と通信オーバーヘッドが膨大になる可能性がある.本論文では,この擬似スペクトル法を用いた乱流解析プログラムを並列化するために,通信時間と通信オーバーヘッドを考慮し,複数の配列を処理する際に,一つの配列がなるべく少数のプロセッサに割り当てられるように分割し,なおかつ,複数の配列の演算が同時処理で, May 2003, 44, SIG6(ACS1), 45-54
  • Refereed, 情報処理学会MPSシンポジウム, GA を用いたレイリー波分散曲線に基づく水平成層構造の推定, TAKATA Masami, Jan. 2003, 2003, 2, 229-231
  • Refereed, 情報処理学会MPSシンポジウム, i アプリを用いた分散 GA の実装, TAKATA Masami, Jan. 2003, 2003, 2, 251-258
  • Refereed, 情報処理学会HPCシンポジウム, 擬似スペクトル法を用いた乱流場の直接数値シミュレーションの並列化と性能評価, TAKATA Masami, Jan. 2003, 2003, 4, 101-108
  • Refereed, In Proceedings of International Conference on Parallel and Distributed Processing Techniques and Applications, A Quantum Algorithm for Searching Web Communities, TAKATA Masami; Mayumi Oto, Jun. 2002, I, 260-266
  • Refereed, In Proceedings of International Conference on Parallel and Distributed Processing Techniques and Applications, Accelerated Program Partitioning Algorithm - An Improvement of Girkar's Algorithm -, TAKATA Masami; Masami Takata, In this paper, we propose several heuristics that improve the branch and bound based program partitioning algorithm proposed by Girkar et. al., and indicate the effectiveness by experiments. The heuristic depends on the element order of edges of given task graphs deeply. Therefor, it is needed to sort edges pertinently for the effective use of the heuristic. The ways of sorting are investigated with experiments. As the results of the experiments, it turned out that approximate solutions, which provide enough practical partitionings, by accelerated heuristic is obtained, and execution times and errors to optimal solutions decrease considerably by sorting edges., Jun. 2002, II, 699-705
  • Refereed, The Proceedings of the 5th International Conference and Exhibition on High-Performance Computing in the Asia-Pacific Region, Effective Parallelization of a Turbulent Flow Simulation, TAKATA Masami; Masami Takata, In this paper, we present parallelization methods for a direct numerical simulation using the Message Passing Interface. The programs has been parallelized for four and eight processors configurations, and we show that the execution time of the parallelized programs for two configurations of workstation clusters, four and eight nodes, are reduced to about 1/4 and 1/8 of the running time of the original sequential program, respectively., Sep. 2001, CD-ROM Proceeding
  • Refereed, In Proceedings of International Conference on Parallel and Distributed Processing Techniques and Applications, An Improvement of Program Partitioning Based Genetic Algorithm, TAKATA Masami; Masami Takata, We propose a sorting rule that improves the genetic algorithm based program partitioning algorithm, and evaluate the effectiveness by experiments. The sorting rule is sensitized the order of nodes of a given task graph. Therefore, it is necessary to change the node number to make effective use of the sorting rule. Several variations of the method are investigated and experimentally evaluated. Approximate solutions that provide a sufficient practical partitioning are obtained using the accelerated sorting method, and execution times and error decreased considerably by changing node numbers of the task graph., Jun. 2000, I, 215-221
  • Refereed, INNOVATIVE ARCHITECTURE FOR FUTURE GENERATION HIGH-PERFORMANCE PROCESSORS AND SYSTEMS, IEEE COMPUTER SOC, A heuristic approach to improve a branch and bound based program partitioning algorithm, M Takata; Y Kunieda; K Joe, In this paper we propose several heuristics that improve the branch and bound based program partitioning algorithm proposed by Girkar et. al., and evaluate the effectiveness by experiments. The heuristic depends heavily on the element order of edges of a given task graph. Therefore, it is necessary to sort the edges carefully to make effective use of the heuristic. Different sorting methods are investigated and experimentally evaluated. Approximate solutions that provide a sufficient practical partitioning were obtained using the accelerated heuristic, and execution times and error compared to the optimal solutions decreased considerably by sorting the edges of the task graph., 2000, 105, 114, International conference proceedings
  • Refereed, Proceedings of the Innovative Architecture for Future Generation High-Performance Processors and Systems, IEEE Computer Society, A heuristic approach to improve a branch and bound based - Program partitioning algorithm, Masami Takata; Yoshitoshi Kunieda; Kazuki Joe, In this paper, we propose several heuristics that improve the branch and hound based program partitioning algorithm proposed by Girkar et. al, and evaluate the effectiveness by experiments. The heuristic depends heavily on the element order of edges of a given task graph. Therefore, it is necessary to sort the edges carefully to make effective use of the heuristic. Different sorting methods are investigated and experimentally evaluated. Approximate solutions that provide a sufficient practical partitioning were obtained using the accelerated heuristic, and execution times and error compared to the optimal solutions decreased considerably by sorting the edges of the task graph., 1999, 1999-, 105, 113, International conference proceedings

MISC

  • Not Refereed, 研究報告数理モデル化と問題解決(MPS), Proposal of Origami Support Tool using AR, Yukino Kato; Masami Takata, Mar. 2022, 2022-MPS-137, 15, 1, 6, Summary national conference
  • Not Refereed, 研究報告数理モデル化と問題解決(MPS), A CRAFT-Based Layout Analysis Method for Early-Modern Japanese Books, 飯田 紗也香; 竹本 有紀; 石川 由羽; 髙田 雅美; 城 和貴, Mar. 2022, 2022-MPS-137, 11, 1, 6, Summary national conference
  • Not Refereed, 研究報告数理モデル化と問題解決(MPS, A Study of an Essential Oil Blending Method Using Q-learning, Moeka Sakamoto; Masami Takata, Dec. 2021, 2021-MPS-136, 16, 1, 4, Summary national conference
  • Not Refereed, 研究報告数理モデル化と問題解決(MPS), 近代書籍文字認識に有効なデータ拡張の一手法, 倉田 帆風; 藤﨑 菜々美; 飯田 紗也香; 高田 雅美; 城 和貴, Dec. 2021, 2021-MPS-136, 10, 1, 6, Summary national conference
  • Not Refereed, 研究報告数理モデル化と問題解決(MPS), Examination of Mutual Translation between Modern Literary Style and Modern Colloquial Style using Unsupervised Learning, Chikako Fujii; Yuki Takemoto; Yu Ishikawa; Masami Takata; Kazuki Joe, Dec. 2021, 2021-MPS-136, 9, 1, 6, Summary national conference
  • Not Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 近代書籍における文字切り出し手法の検討, 飯田 紗也香; 竹本 有紀; 石川 由羽; 髙田 雅美; 城 和貴, Feb. 2021, 2020-MPS-132, 4, 1, 6
  • 情報処理学会研究報告数理モデル化と問題解決(MPS), 観光スケジュール探索アルゴリズムへの遺伝的アルゴリズムの適用, 和田 直子; 高田 雅美, Dec. 2020, 2020-MPS-131, 16, 1, 6
  • Not Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 邦字新聞における近代文語体と現代口語体の自動翻訳の検討, 稲見 郁乃; 竹本 有紀; 石川 由羽; 高田 雅美; 上田 薫; 城 和貴, Dec. 2020, 2020-MPS-131, 13, 1, 6
  • Not Refereed, 日本応用数理学会 2021年 研究部会連合発表会, デコンボリューション問題におけるcondition L-curveの検討, 久保井 五貴; 田中 利佳; 小澤 伸也; 細田 陽介; 高田 雅美; 木村 欣司, Mar. 2021, Summary national conference
  • IPSJ SIG technical reports, Information Processing Society of Japan (IPSJ), Development of real time respiration monitoring system by using standing wave radars, 高木 里穂; 石川 由羽; 高田 雅美; 梅田 智弘; 城 和貴, Sudden infant death syndrome is a disease that a well-looking baby of up to one year stops his/her breathing without any reason, and comes to die. To prevent babies from this disease, some kind of systems that check baby's breathing circulating system to detect abnormal foretaste is required. Such error detection for baby's breathing circulating can be realized by standing wave radars, which measures the distance of target objects precisely. Measuring series of displacements to the breast of a baby by using standing wave radars. we apply Fourier transform analysis for the frequency band of baby's breathing, and we present the development of respiration monitoring system to observe of baby's breathing., 19 Jan. 2015, 2015, 1, 1, 6
  • IPSJ SIG Notes, Information Processing Society of Japan (IPSJ), Development of real time respiration monitoring system by using standing wave radars, 高木 里穂; 石川 由羽; 高田 雅美; 梅田 智弘; 城 和貴, Sudden infant death syndrome is a disease that a well-looking baby of up to one year stops his/her breathing without any reason, and comes to die. To prevent babies from this disease, some kind of systems that check baby's breathing circulating system to detect abnormal foretaste is required. Such error detection for baby's breathing circulating can be realized by standing wave radars, which measures the distance of target objects precisely. Measuring series of displacements to the breast of a baby by using standing wave radars. we apply Fourier transform analysis for the frequency band of baby's breathing, and we present the development of respiration monitoring system to observe of baby's breathing., 19 Jan. 2015, 2015, 1, 1, 6
  • IPSJ SIG Notes, Information Processing Society of Japan (IPSJ), Partial recognition of three-dimensional object In volume data, Eri Nagao; Yu Ishikawa; Masami Takata; Joe Kazuki, In this paper, we propose a method of partial recognition for three dimensional object in volume data. In hospitals, MRI (magnetic resonance imaging) and CT (Computed Tomography), which take photos as slice images, are often adapted to obtain information in inner of body. By constructing slice images to a three-dimensional object, users understand information of slice images intuitively. By an automatic partial recognition, location of pathology can be clear speedy. In the proposed method, slice images are transformed into volume data and a partial recognition is performed using LVQ (Learning Vector Quantization) with the three-dimentional SIFT features., 02 Dec. 2014, 2014, 9, 1, 6
  • 電子情報通信学会技術研究報告 = IEICE technical report : 信学技報, The Institute of Electronics, Information and Communication Engineers, Ruby Removal Filters by Genetic Programming using the classification of printing type data for Early-Modern Japanese Printed Books, 粟津 妙華; 高田 雅美; 城 和貴, In National Diet Library, books which are possessed in library as "the digital library from meiji era" are open to the public on Web. Since these are shown as image data and cannot search using document contents, an automatic text conversion is needed. There is a major obstacle to text conversion. It is ruby. Ruby can not be removed in the histogram method. Therefore, we have proposed a ruby removal method for early-modern Japanese printed books. However, since the proposed method is based on the external information added to the books, the feasibility is low. In this paper, we propose a method to remove the ruby automatically from early-modern Japanese printed books by generating ruby removal formula in Genetic Programming using the training data was based on the data of book image., 25 Jun. 2014, 114, 105, 117, 122
  • 研究報告ハイパフォーマンスコンピューティング(HPC), 一般社団法人情報処理学会, 疎行列ベクトル積性能を決める諸要因, 田邊昇; 冨森苑子; 高田雅美; 城和貴, 疎行列ベクトル積 (SpMV) は多くの場合にキャッシュアーキテクチャとの相性が悪い.並列処理においては負荷不均衡が性能に与える影響も大きい.これまでは SpMV 性能を決める要因として,キャッシュのヒット率や一行あたりの非零要素数の平均,最大値,分散が注目されていた.しかし,それらと性能との相関が不明瞭であり,SpMV の挙動は長年にわたり謎に包まれていた.それは SpMV の最適化や,効率的な疎行列ライブラリ構築の障害であった.本報告では,SpMV 性能を左右する様々な要因をアプリケーション依存の要因とプラットフォーム依存の要因に分けて考察した.それを踏まえて行列の非零要素配置から導かれる時間的局所性と空間的局所性等のアプリ依存パラメータを導入した SpMV 性能モデルを構築した.その上でフロリダ大コレクションから抜粋した 115 種の疎行列と GPU を用いて SpMV 性能モデルの評価実験を行った.その結果,GPU 上で実行する場合は Padding に関する補正と小さな行列での補正が必要であることと,長行を折り畳むなど適切な負荷分散がなされた場合はキャッシュのヒット率よりも,空間的局所性やインデックス転送の抑制の方が実効性能に敏感であることが明らかになった., 24 Feb. 2014, 2014, 7, 1, 10
  • 研究報告ハイパフォーマンスコンピューティング(HPC), 一般社団法人情報処理学会, アーキテクチャの進歩を促進するGraph500ベンチマークのあり方, 田邊昇; 冨森苑子; 高田雅美; 城和貴, Graph500 はビッグデータ解析や疎行列処理のベンチマークとも言われており,Top500 を補うものとしても近年注目を集めてきた.従来の Graph500 参照実装は従来の計算機が苦手とする疎行列ベクトル積系応用全般を代表し,計算機の発展を促進する力があった.一方,ビットマップ式訪問管理と Hybrid BFS アルゴリズムがデファクトになる流れにある.筆者らの予備調査の結果,それらを実装した Pre release 版 Graph500 はキャッシュに適合し,メモリアクセス負荷が大幅に軽減され,本来とは異質なワークロードになることが明らかになってきた.本報告では,計算機の発展を促進する観点から,スーパーコンピュータのランキングに利用されるベンチマークの役割と,その望ましいあり方を考察する.さらに Hybrid 動作における局所性とキャッシュ挙動の観点から上記の予備調査結果を補強するとともに,Graph500 の将来について考察する., 23 Sep. 2013, 2013, 18, 1, 9
  • 年会論文集, 日本教育情報学会, 10J05 教材開発(奥行き学習教材(特許第5226821号))の報告と今後の課題(タブレット端末と教材開発,一般研究), 吉井 直子; Peterson Ward; 高田 雅美; 吉住 秀; 城 和貴, 算数・数学で学習する幾何分野,特に空間図形(3次元物体)の理解を高めるための教材開発・教材研究を行っている.初めに奥行き知覚の実験を行い,得られた実験データからオクリュージョン(重なり)効果を検証した.そこでオクリュージョンを用いた「奥行き学習教材」を開発し,2013年3月に特許権を取得した.本稿では特許発明の過程を報告し,教材の活用方法を紹介する.また知的財産権について述べる., 2013, 29, 406, 407
  • Not Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 地震予知を目的としたDEMETER衛星データの雷ノイズ除去, 石通彩加; 竹本有紀; 石川由羽; 髙田雅美; 城和貴, Feb. 2020, 2020-MPS-127, 19, 1, 6
  • Not Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 近代書籍における低出現頻度文字種の獲得, 藤田未希; 竹本有紀; 石川由羽; 髙田雅美; 城和貴, Dec. 2019, 2019-MPS-126, 6, 1, 6
  • Not Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 近代文語体と現代口語体の自動翻訳への試み, TAKATA Masami, Dec. 2018, 2018-MPS-121, 18, 1-6
  • Not Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 帝国議会会議録におけるレイアウト解析手法の比較, TAKATA Masami, Sep. 2018, 2018-MPS-120, 13, 1-6
  • Not Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), シフト付きCholesky QR分解を用いた特異値分解の高速化, TAKATA Masami, Dec. 2017, 2017-MPS-116, 1, 1-6
  • Not Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 透過型HMD上での日本語文字切り出しの一手法, TAKATA Masami, Feb. 2017, 2017-MPS-112, 7, 1-6
  • Not Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 計測位置のずれを考慮した脳波個人認証の検証, TAKATA Masami, Feb. 2017, 2017-MPS-112, 8, 1-6
  • Not Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 近代書籍用フォントの自動生成, TAKATA Masami, Feb. 2017, 2017-MPS-112, 15, 1-6
  • Not Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), DCGANを用いたイラスト画像生成の一手法, TAKATA Masami, Feb. 2017, 2017-MPS-112, 16, 1-6
  • Not Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 調剤薬局用薬剤売買のためのレコメンドシステム, TAKATA Masami, Dec. 2016, 2016-MPS-111, 7, 1-5
  • Not Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), Myoを用いた腕の表面筋電位解析, TAKATA Masami, Sep. 2016, 2016-MPS-110, 9, 1-6
  • Not Refereed, 情報処理学会研究報告数理モデル化と問題解決(MPS), 遺伝的アルゴリズムを用いたQWOPの運動学習の一手法, TAKATA Masami, Sep. 2016, 2016-MPS-110, 10, 1-6
  • Not Refereed, 情報処理学会研究報告, Deep Learningを用いたSNSからの知識抽出の一手法, TAKATA Masami, Mar. 2016, 2016-MPS-107, 1, 1-6
  • Not Refereed, 情報処理学会研究報告, 脳波計測位置のずれが脳波個人認証に与える影響, TAKATA Masami, Sep. 2015, 2015-MPS-105, 9, 1-6
  • Not Refereed, 情報処理学会研究報告, 近代書籍用OCRのための学習用特定フォントセットの自動生成手法, TAKATA Masami, Sep. 2015, 2015-MPS-105, 10, 1-6
  • Not Refereed, 情報処理学会研究報告, 上肢神経系セルフチェックのためのHapLogを用いた評価システムの試作, TAKATA Masami, Sep. 2015, 2015-MPS-105, 14, 1-6
  • Not Refereed, 情報処理学会研究報告, 呼吸変動解析によるストレス指標の提案, TAKATA Masami, Jun. 2015, 2015-MPS-103, 22, 1-6
  • Not Refereed, 情報処理学会研究報告, COG分類を考慮した遺伝子発現プロファイルデータのクラスタリング手法, TAKATA Masami, Jun. 2015, 2015-MPS-103, 23, 1-6
  • Not Refereed, 電子情報通信学会技術研究報告 信学技報, 心拍波形の振幅と面積を用いた個人認証の提案, TAKATA Masami, Jun. 2015, 115, 117, 11-16
  • Not Refereed, 情報処理学会研究報告, Leap Motion Controllerを用いた指文字認識, TAKATA Masami, Feb. 2015, 2015-MPS-102, 8, 1-6
  • Not Refereed, 情報処理学会研究報告, ニューラルネットワークを用いた心拍データの学習と利用, TAKATA Masami, Feb. 2015, 2015-MPS-102, 17, 1-6
  • Not Refereed, 情報処理学会研究報告, ボリュームデータにおける3次元物体の部分認識, TAKATA Masami, Dec. 2014, 2014-MPS-101, 9, 1-6
  • Not Refereed, 情報処理学会研究報告, 機械学習を用いた問題解答のための推論システムの開発, TAKATA Masami, Dec. 2014, 2014-MPS-101, 10, 1-6
  • Not Refereed, 第34回セマンティックウェブとオントロジー研究会, 表構造を用いた漢方薬オントロジー自動構築手法の提案, TAKATA Masami, Nov. 2014, SIG-SWO-A1402-06
  • Not Refereed, 第19回日本バーチャルリアリティ学会大会, プリシェイピングを用いたVRユーザのモーション認識, TAKATA Masami, Sep. 2014, CD, 419-422
  • Not Refereed, 第19回日本バーチャルリアリティ学会大会, VR用体感アミューズメントアプリの開発, TAKATA Masami, Sep. 2014, CD, 576-579
  • Not Refereed, 情報処理学会研究報告, 活字データの分類を用いた進化計算による近代書籍からのルビ除去, TAKATA Masami, Jun. 2014, 2014-MPS-98, 20, 1-6
  • Not Refereed, 人工知能学会研究会資料, Development of a Physical Activity Recommendation System for Smart Health-Care, TAKATA Masami, Mar. 2014, SIG-SWO-A1303-02, 1-8
  • Not Refereed, 人工知能学会研究会資料, An Automatic Ontology Concepts Extension Method, TAKATA Masami, Mar. 2014, SIG-SWO-A1303-05, 1-8
  • Not Refereed, IPSJ SIG Notes, Information Processing Society of Japan (IPSJ), Modeling of physical activity including Non-Exercise Activity Thermogenesis using plural triaxial accelerometers, TAKATA Masami, The metabolism by physical activity of low exercise intensity in daily life (Non-Exercise Activity Thermogenesis: NEAT), account for 20-30% of total energy expenditure of the people, its individual difference is large. Therefore, we propose the modeling for estimating energy expenditure of any physical activities including not only exercise but also NEAT. Energy expenditure is in proportion to body weight. Accordingly we model by limiting the energy expenditure of the body having a large specific gravity in the whole body. To capture more accurate movement of the body, we use three-axis acceleration data of the three points, right thoracic and left thoracic and lumbar. We propose a model of energy expenditure by physical activity including NEAT from momentum, kinetic energy, impulse, and work., Feb. 2014, 2014-MPS-97, 21, 1-6, 6
  • Not Refereed, IPSJ SIG Notes, Information Processing Society of Japan (IPSJ), 多フォント漢字認識手法における各カテゴリと必要教師データ数の分析, TAKATA Masami, Since the fonts of the Kanji character in early-modern Japanese printed books dating from the Meiji era to the first half of Showa era (AD1868-1945) has the greatly different feature from the standard fonts determined by JIS. We have been proposed the multi-fonts Kanji character recognition method suitable for early-modern Japanese printed books. In this paper, the multi-fonts Kanji character recognition method is applied to 1000 kinds of Kanji characters, and the relationship the between the feature of Kanji characters and the numbers of necessary training data is analyzed. In our proposed method, when Kanji characters are recognized with the PDC feature and SVM, the training data is needed. However, it was not analyzing about the relation between Kanji characters and the number of necessary training data. Therefore, the reration between the feature of Kanji characters and the number of necessary training data is analyzed using the feature value of Kanji characters by PDC vector and the change of recognition rate for each category of training data., Feb. 2014, 2014-MPS-97, 10, 1-6, 6
  • Not Refereed, IPSJ SIG Notes, Information Processing Society of Japan (IPSJ), オントロジーを用いた建造物のバリアフリー度判定, TAKATA Masami, In this research, we construct a 3D surface model, which is translated in information of standing wave radar. The standing wave radar can measure distance to an object without contact. In the information of the standing wave radar, an optimal combination algorithm is adopted. Then, once the three-dimensional information is transformed into the two-dimensional coordinates, a 3D surface model is constructed by using the Delaunay triangulation., Feb. 2014, 2014-MPS-97, 15, 1-6, 6
  • Not Refereed, 情報処理学会研究報告, 定在波レーダを用いた距離画像の構築, TAKATA Masami, Feb. 2014, 2014-MPS-97, 19, 1-6
  • Not Refereed, 情報処理学会研究報告, 脳波個人認証のための特徴抽出, TAKATA Masami, Feb. 2014, 2014-MPS-97, 20, 1-6
  • Refereed, Proceedings of The 2014 International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA2014), Acceleration of numerical computation for positive roots of polynomials by improved bounds, Kinji Kimura; Takuto Akiyama; Hiroyuki Ishigami; Masami Takata; Yoshimasa Nakamura, 2014, II, 201, 207, Summary international conference
  • Not Refereed, 情報処理学会研究報告\nグラフィックスとCAD, 3次元形状データにおける色段差を利用した特徴抽出法の提案, TAKATA Masami, Nov. 2013, 2013-CG-153, 20, 1-6
  • Not Refereed, IPSJ SIG Notes, Information Processing Society of Japan (IPSJ), Graph500のHybrid解法に内在する局所性, TAKATA Masami, Graph500 is a benchmark suite for big data analysis and sparse matrix processing which receives attention in these years. Since the use of a bitmap to mark the visit of vertices and Beamer's Hybrid BFS algorithm to make a part of a sparse matrix be accessed are reported, we believe that such methods are to be a de facto standard. In this paper, from the viewpoint of reference locality, we analyze the memory access characteristics of pre-release version implemented with OpenMP and the two methods for the latest Graph500 reference code. As the results, we confirm that the number of memory accesses, a spatial locality metric of row index that shows performance degradation in the case of cache misses, and temporal locality (average working set sizes) that shows the resistance to cache misses, change drastically. We confirmed that the reduction rate of the number of memory accesses and temporal locality are accelerated in proportion to the target graph sizes. It means that the Graph500 with Hybrid method is another benchmark suite to measure the characteristics of optimized BFS implementation, while the conventional Graph500 represents most applications with sparse matrix vector multiplications that are not efficiently performed on conventional computer systems to require the improvement of computer systems. Furthermore, the hot spots against processing times do not include the functions to perform BFS, and we find it is no longer the benchmark for search speed practically. To investigate the essential meaning of the benchmarks, we propose appropriate and practical rules that do not damage it., Jul. 2013, 2013-HPC-140, 27, 1-8, 8
  • Not Refereed, 情報処理学会研究報告, Scatter/Gatherをどこで行なうべきか?-低電力で高スループットなランダムアクセスのために-, TAKATA Masami, HPC における反復解法のみならず、近年ではビッグデータ解析のニーズの高まりにより、大量データへのランダムアクセスの低電力化・高スループット化のニーズが高まっている。キャッシュベースの CPU や GPU において、キャッシュから溢れる配列に対して Scatter/Gather を行なうと、ライン内の空間的局所性の欠乏により消費電力とスループットの両面で深刻な問題が発生する。特に電力やメモリスループットが不足する将来の大規模計算基盤においては、この問題が年々深刻さを増す。本報告では、上記の問題の解決策として、Scatter/Gather 機能を Hybrid Memory Cube 内で行なうことを提案する。提案方式の電力やスループットに関するメリットについて、モデルを構築しつつ、Graph500 ベンチマーク課題行列に対する疎行列ベクトル積を実例に考察する。, Mar. 2013, 2013-ARC-204, 4, 1-7, 7
  • Not Refereed, 情報処理学会研究報告, 疎行列キャッシュ適合性に基づくGraph500 ベンチマークの特性解析, TAKATA Masami, Graph500 はビッグデータ解析や疎行列処理のベンチマークとも言われており,近年注目を集めている.本論文では Graph500 の疎行列や,その Randomize や先行研究の Vertex sorting の有無を制御した行列のキャッシュメモリへの適合性を,空間的局所性の観点から解析した.その結果,Graph500 の疎行列の空間的局所性は極めて低く,疎行列ベクトル積 (SpMV) を行う場合にはベクトルのアクセスの際に,キャッシュラインには平均して 1 個程度しか有効なデータは載っていないことが判った.ランダマイズによりその傾向が強まる.Vertex sorting は空間的局所性にはあまり作用しなかった.上記行列の SpMV 実行時の GPU の L1 キャッシュヒット率は,Vertex sorting の効果が小さく,グラフサイズ (SCALE) が大きくなるほど低下した.データサイズの大きさと非零要素配置のランダム性のため,キャッシュアウェアなソフトウェア最適化が困難である.特に空間的局所性の低さに伴う問題は現時点では未開拓な性能改善要因である.ゆえに,Graph500 や Green Graph500 の競争で優位に立つには,1 桁程度のメモリトラフィック削減による高速化や省電力化が期待できるメモリ側の Gather 機能が有望である.Graph500 is a benchmark suite for big data analysis and sparse matrix processing which receives attention in these years. The spatial locality of sparse matrices used for Graph500 and their variations which are controlled in adoption of randomization and vertex sorting which is a technique of preceding work are investigated. We show the spatial locality of sparse matrices used for Graph500 is very low and there is about 1 or a little more valid data on a cache line for the memory accesses issued by sparse matrix-vector multiplications (SpMV) in average. The tendency is increased by randomization. The effect of vertex sorting does not have big effects on spatial locality. Effect of vertex sorting on L1 hit rate of GPU on processing SpMV for the above matrices is small and is degraded when the size of graph (i.e. SCALE) becomes larger. It is very difficult to solve the problem by just software approach because of the huge size of sparse matrices and the randomness of their accesses to degrade the optimization against their cache awareness. Especially problems caused by low spatial locality are frontiers of improvement, now. Therefore, hardwired gather functions at memory side is promising for taking advantage in the Graph500 lists, which improve the processing speed due to reduction of memory traffic in an order of magnitude., Feb. 2013, 2013-HPC-138, 30, 1-7, 7
  • Not Refereed, 情報処理学会研究報告, 時間的局所性を考慮した疎行列のキャッシュ適合性, TAKATA Masami, Dec. 2012, 2012-HPC-137, 28, 1-8
  • Not Refereed, 情報処理学会研究報告, LF/HFを用いた時空間ストレス指標の提案, TAKATA Masami, Dec. 2012, 2012-MPS-91, 2, 1-6
  • Not Refereed, 情報処理学会研究報告, 遺伝的プログラミングを用いた近代書籍からのルビ除去, TAKATA Masami, Dec. 2012, 2012-MPS-91, 17, 1-6
  • Not Refereed, 情報処理学会研究報告, 定在波レーダを用いた呼吸モニタリングシステムの開発, TAKATA Masami, Dec. 2012, 2012-MPS-91, 19, 1-6
  • Not Refereed, 情報処理学会研究報告, 書籍オントロジーのための推論部の開発, TAKATA Masami, Dec. 2012, 2012-MPS-91, 32, 1-6
  • Not Refereed, 情報処理学会研究報告, α波構成要素の切出しと解析, TAKATA Masami, Dec. 2012, 2012-MPS-91, 34, 1-6
  • Not Refereed, 電子情報通信学会技術研究報告, グラフ解析ワークロードのキャッシュ適合性, TAKATA Masami, Oct. 2012, 112, 237, 66-72
  • Not Refereed, 情報処理学会研究報告, 視覚障害者のための墨字学習支援システムの開発と評価, TAKATA Masami, Sep. 2012, 2012-MPS-90, 1, 1-6
  • Not Refereed, 情報処理学会研究報告, 古典油彩画グレーズ技法シミュレーションの高速化, TAKATA Masami, コンピュータ上での絵具の発色シミュレーションは,グラフィックソフトウェアへの応用が期待できる.古典油彩画における 「グレーズ」 は,薄く溶いた透明な絵具の層を重ねる技法であり,混色で表現できない色調を,重色によって表現するものである.その発色は,放射伝達方程式を解くことによって求められる.しかし従来手法は,異なる絵具でグレーズを行う方法が示されておらず,高精度な解を得るには計算量が大きくなるため,グラフィックソフトウェアへの応用には不向きである.本研究では,まず従来手法を異なる絵具の層に対応できるように拡張する.さらに,係数行列を分解し,刻み幅を固定することにより,高精度な解を小さい計算量で得ることが可能な数値解法を提案する.また,実際の油絵具のデータを用いて数値実験を行い,高精度な解が,従来手法の約 4 % ~ 7 %の計算時間で求められることを示す., Sep. 2012, 2012-MPS-90, 7, 1-6, 6
  • Not Refereed, 情報処理学会研究報告, 同一出版者の近代書籍に対する漢字認識評価, TAKATA Masami, Sep. 2012, 2012-MPS-90, 25, 1-6
  • Not Refereed, 情報処理学会研究報告, 疎行列のキャッシュへの適合性分類に関する予備評価, TAKATA Masami, Jul. 2012, 2012-HPC-135, 17, 1-7
  • Not Refereed, 情報処理学会研究報告, 近代デジタルライブラリーの文字切り出しにおける実際的手法, TAKATA Masami, Feb. 2012, 2012-MPS-87, 34, 1-6
  • Not Refereed, 情報処理学会研究報告, ARを用いた空間認識能力向上のための学習方法, TAKATA Masami, Feb. 2012, 2012-MPS-87, 33, 1-6
  • Not Refereed, 5TH BIOMEDICAL ENGINEERING INTERNATIONAL CONFERENCE (BMEICON 2012), IEEE, Constitution and Phase Analysis of Alpha Waves, Yu Ishikawa; Masami Takata; Kazuki Joe, 2012, 98, 98, Summary international conference
  • Not Refereed, 情報処理学会研究報告, Enobioを用いたα波形状の推定, TAKATA Masami, Dec. 2011, 2011-MPS-86, 27, 1-6
  • Not Refereed, 情報処理学会研究報告, 未病の認識・改善に有効なストレスオントロジーの設計, TAKATA Masami, Dec. 2011, 2011-MPS-86, 26, 1-6
  • Not Refereed, 情報処理学会研究報告, MKLによる食事画像認識の追試, TAKATA Masami, Dec. 2011, 2011-MPS-86, 17, 1-6
  • Not Refereed, 情報処理学会研究報告, 異なるプラットフォームで効果的な3Dモデルを表示するためのポリゴンリダクションの方法, TAKATA Masami, Dec. 2011, 2011-MPS-86, 16, 1-6
  • Not Refereed, 情報処理学会研究報告, 不規則型応用を加速するメモリアクセラレータ-Exa FLOPSマシンの文脈から, TAKATA Masami, Nov. 2011, 2011-HPC-132, 14, 1-8
  • Not Refereed, 情報処理学会研究報告, e-口コミを活用したe-learning教材の開発とその評価, TAKATA Masami, Sep. 2011, 2011-MPS-85, 27, 1-6
  • Not Refereed, 情報処理学会研究報告, メモリアクセラレータで強化したGPUのCG法による評価, TAKATA Masami, Jul. 2011, 2011-HPC-130, 15, 1-10
  • Not Refereed, 情報処理学会研究報告, データ通信時間を隠蔽したGPGPUによる粒子ベースボリュームレンダリングの高速化, TAKATA Masami, Mar. 2011, 2011-MPS-82, 22, 1-6
  • Not Refereed, 情報処理学会研究報告, GPUを用いた複数顔検出の高速化に関する一手法, TAKATA Masami, Mar. 2011, 2011-MPS-82, 21, 1-6
  • Not Refereed, 情報処理学会研究報告, SVMに基づく多フォント漢字認識手法の評価, TAKATA Masami, Mar. 2011, 2011-MPS-82, 14, 1-6
  • Not Refereed, 情報処理学会研究報告, 近代デジタルライブラリーテキスト化支援のためのポータルサイトの設計, TAKATA Masami, Dec. 2010, 2010-MPS-81, 35, 1-6
  • Not Refereed, 情報処理学会研究報告, CUDAに適用したループ再構築手法の評価, TAKATA Masami, Sep. 2010, 2010-MPS-80, 6, 1-6
  • Not Refereed, 情報処理学会研究報告, 情報処理学会, 機能メモリとGPUのPCI express接続によるヘテロ環境における超大規模疎行列ベクトル積の性能予測, TAKATA Masami, 本報告では,疎行列ベクトル積のベクトルがデバイスメモリに入りきらないほど大きな問題向けに,GPU が PCI express 越しに Gather 機能を有する大容量機能メモリをアクセスするシステムを用いた並列処理方式を提案する.フロリダ大の疎行列コレクションを用いて提案方式の性能評価を行った.その結果,間接アクセスの直接アクセス化により,単体性能は既存研究の最大 4.1 倍に向上した.GPU 内キャッシュが溢れる心配も無い.GPU 間通信を完全に排除可能にした構成によりスケーラビリティは保証されており,PCI express のバースト転送バンド幅で制約される単体性能にノード数を乗じたものが並列実効性能となる.In this report, we propose a parallel processing strategy for huge scale sparse matrix-vector product whose vector cannot be held on a device memory. The strategy uses a system connected by PCI express with GPUs and functional memories with gather function. We evaluate the performance of proposed strategy with University of Florida Sparse Matrix Collection. The result shows the 4.1 times acceleration over the existing performance record with a GPU in the maximum case. There is no risk of performance degradation by overflowing cache capacity on GPU. Because of the architecture without inter-GPU communications, scalability is guaranteed. Therefore, parallel effective performance is the product of number of nodes and single GPU performance limited by burst transfer bandwidth of PCI express., Jul. 2010, 2010-HPC-126, 20, 1-9, 9
  • Not Refereed, 情報処理学会研究報告, タンパク質立体構造の動きに着目した時系列解析, TAKATA Masami, Mar. 2010, 2010-BIO-20, 5, 1-6
  • Not Refereed, 情報処理学会研究報告, 反射を考慮した日照量計算手法の評価, TAKATA Masami, Mar. 2010, 2010-MPS-77, 36, 1-6
  • Not Refereed, 情報処理学会研究報告, 収集方法の異なるウェブアーカイブの相互利用, TAKATA Masami, Mar. 2010, 2010-MPS-77, 32, 1-6
  • Not Refereed, 情報処理学会研究報告, リアルタイムステレオQueryballの開発, TAKATA Masami, Mar. 2010, 2010-MPS-77, 31, 1-6
  • Not Refereed, 情報処理学会研究報告, ビデオシーケンスを利用したVR環境マッピング用テクスチャ及び空間の生成, TAKATA Masami, Mar. 2010, 2010-MPS-77, 26, 1-6
  • Refereed, Proceedings of The 2010 International Conference on Parallel and Distributed Processing Techniques and Applications(PDPTA2010), Vol.II, 2010, pp.619–625, 情報処理学会, Generating algorithms for matrices with large condition number to evaluate singular value decomposition, Masami Takata; Kinji Kimura; Yoshimasa Nakamura, 本論文では,特異値分解を評価するために,条件数の大きなテスト行列の作成法を提案する.我々が対象とする条件数は,以下の 2 種類である.1 つ目は,連立 1 次方程式を解く際の困難さを 1 つの指標とする.2 つ目は,特異値の近接度を用いる.1 つ目の提案作成法では,2 重対角行列のみならず,密行列を作成することも可能である.一方,2 つ目の提案作成法では,2 重対角行列のみが作成可能である.提案する 2 種類の作成法の目的は異なるため,それぞれに意義がある.これらの作成法を用いて,LAPACK 3.2.1 に含まれているいくつかの特異値分解アルゴリズムを評価する.In this paper, we propose new generating algorithms for matrices with large condition number to evaluate singular value decomposition. We target two types of condition number. The first means intractableness to solve simultaneous linear equation. The second uses adjacent amount in each singular value. The first proposed algorithm can generate not only bidiagonal but also dense test matrices. On the other hand, the second proposed algorithm can generate only bidiagonal test matrices. Since targets in the proposed algorithms are different, it is important to generate test matrices in two types of condition number. By using two generating algorithms, some singular value decomposition routines in LAPACK version 3.2.1 are evaluated., 2010, 2010, 1, 1, 6
  • Not Refereed, 情報処理学会研究報告, 情報処理学会, 機能メモリとCOTSのPCI express接続によるヘテロ環境向けボリュームレンダリングの設計, TAKATA Masami, 本報告では,筆者らが提案している可視化システムにおける 「アクセラレータが PCI express 越しに機能メモリをアクセスする」 という概念を推進するために,ボリュームレンダリングに注目した提案システム向けアルゴリズムを検討し,PCI express の特性を評価する.さらに,その環境に適したボリュームレンダリングを C-ray をベースに設計する. C-ray は Cell/B.E. 用に最適化されたボリュームレンダリングの実装例であり,我々の開発しているシステムは PPE が入っていないので,C-ray をそのまま用いることはできない.そのための移植手法の検討が本稿の目的である.In this report, aiming at our concept of "Accelerators access function memory via PCI express ", we investigate a volume rendering algorithm for our system to evaluate the characteristics of PCI-express. In addition, we design a volume rendering algorithm based on the C-ray for the hetero environment. The C-ray is the optimized implementation of volume rendering for the Cell/B.E. Broadband Engine. Since our system does not have PPEs, we can not apply the C-ray to our system. The purpose of this report is to investigate the re-design of the C-ray for our system., Nov. 2009, 2009-HPC-123, 1, 1-6, 6
  • Not Refereed, 情報処理学会研究報告, 対話的実時間遠隔可視化システムの構築, TAKATA Masami, Sep. 2009, 2009-MPS-75, 19, 1-6
  • Not Refereed, 日本教育情報学会年会論文集(25), 日本教育情報学会, 視覚特性(オクリュージョン効果)を利用した教育への試み, TAKATA Masami, 初等中等教育現場では印刷物をよく使用する.ところが,子どもの2次元画像の認識に関する既存研究は少なく,男女別画像別に分析したものは,我々が知る限り存在しない.そこで,奥行き知覚の発達過程について実験を行い考察した.実験結果から,2次元画像の奥行き知覚について男子と女子の間には発達の違いがあること,黒板にかかれた空間図形などのエッジ画像では正解率が低くなることが検証できた.そこで,この問題を補完する一例としてオクリュージョン(重なり)効果を採り上げる., Aug. 2009, 25, 156-159, 159
  • Not Refereed, 情報処理学会研究報告, 情報処理学会, ランダムアクセス型応用のためのPCI express越しに機能メモリをアクセスするアーキテクチャ, TAKATA Masami, 本報告では,筆者らが提案した可視化装置における 「アクセラレータが PCI express 越しに機能メモリをアクセスする」 という概念を,ポストベクトル機向けに整理する.ボリュームレンダリングを例として提案装置向けアルゴリズムを検討した.さらに,本用途への GPU 利用の可能性や PCI express の特性を評価した.乱数やボリュームレンダリングの履歴 index を用いた配列間接参照実験では、ミッドレンジ GPU は理論ピークの 1.3% の実効バンド幅でデバイスメモリをアクセスした.このバンド幅は PCI express と機能メモリにより置き換え可能と考えられる.よって,本概念はランダムアクセス型応用を加速するアクセラレータ向けのメモリ容量向上策や、ECC 付加による信頼性強化策として有望である。In this report, we organize the concept on our visualization machine as for the post-vector supercomputer. The proposed concept is the architecture with accelerators accessing functional memory through PCI express. We investigate the volume rendering algorithm for the machine based on the concept. Moreover, we evaluate the feasibility of using GPU and the performance of PCI express for this use. In the experiment of indirect array accesses using random index and index trace on volume rendering, a mid-range GPU accessed device memory with 1.3% bandwidth compared to the theoretical peak. This bandwidth seemed to be able to be replaced with PCI express and functional memory. Therefore, this concept is promising for accelerator running applications with random accesses both to overcome limitation of device memory capacity, and to improve robustness of memory using ECC., Jul. 2009, 2009-HPC-121, 25, 1-8, 8
  • Not Refereed, 情報処理学会研究報告, タンパク質立体構造における時系列特徴抽出, TAKATA Masami, Mar. 2009, 2009, 19, 97-100
  • Not Refereed, 情報処理学会研究報告MPS-73, 近代書籍に特化した多フォント活字認識法, TAKATA Masami, Mar. 2009, 2009, 19, 205-208
  • Not Refereed, 情報処理学会研究報告MPS-73, 地震短期予測研究のための地電流解析ポータルの開発, TAKATA Masami, Mar. 2009, 2009, 19, 209-212
  • Not Refereed, 情報処理学会研究報告, SVMを用いた生体分子への金属結合部位予測手法の提案, TAKATA Masami, Mar. 2009, 2009, 19, 213-216
  • Not Refereed, 第16 回「ハイパフォーマン スコンピューティングとアーキテクチャの評価」に関する北海道ワークショップ(HOKKE-2009)2009-ARC-174/HPC-119, Cell/B.E.とDIMMnetを併用 した大容量ボリュームレンダリングの並列処理性能, 田邊昇; 佐々木愛美; 中條拓伯; 高田雅美; 城和貴, 2009, -
  • Not Refereed, 情報処理学会研究報告MPS-72, Flashを用いたWebアルバムの提案と設計, TAKATA Masami, Dec. 2008, 2008, 126, 1-4
  • Not Refereed, 情報処理学会研究報告MPS-72, 多目的に衛星データを利用するためのコンテンツ開発環境, TAKATA Masami, Dec. 2008, 2008, 126, 5-8
  • Not Refereed, 情報処理学会研究報告MPS-72, 芝生用塗料の劣化モデルの構築, TAKATA Masami, Dec. 2008, 2008, 126, 17-20
  • Not Refereed, 情報処理学会研究報告MPS-72, Webアーカイブにおける差分収集に用いる更新間隔推定手法の開発とその評価, TAKATA Masami, Dec. 2008, 2008, 126, 41-44
  • Not Refereed, 情報処理学会研究報告MPS-72, 反射を考慮した日照量計算の一手法, TAKATA Masami, Dec. 2008, 2008, 126, 45-48
  • Not Refereed, 電子情報通信学会技術研究報告SP2008-1?19, 書字運動の訓練支援ソフトウェアの開発, TAKATA Masami, May 2008, 108, 66, 79-83
  • Not Refereed, IPSJ SIG Notes, Information Processing Society of Japan (IPSJ), E-Cell Systemを用いた酵母菌のストレス応答MAPK経路のモデル構築, TAKATA Masami, It is supposed that yeasts with alcoholic tolerance are suitable to generate high density alcohol in the process of Nippon Shuzo Brewery. Actually, a lot of compensation including the yeast extinction though various experiments are conducted to discover such yeasts. We try to model the response mechanism of yeast cells to alcohol by computer simulation to reduce such compensation. It pays attention to the relation between alcoholic tolerance and yeast cell wall. In this paper, we model a partial cell wall synthesis Pkc1-MAPK pathway using the E-Cell System., Mar. 2008, 2008, 17, 1-4, 4
  • Not Refereed, IPSJ SIG Notes, Information Processing Society of Japan (IPSJ), リアルタイムQueryballの開発に向けて, TAKATA Masami, Queryball is a tool to understand objects in three-dimensional space. By overlapping a translucent globe with 3D object, it can quiry to the part where the object and the globe come in succession. It can change the display method with visually checking according to various directions. However, the display speed the volume rendering of Queryball to the present stage slows the large size of the object data, and is not for practical use. Therefore, performance improvement by applying GPGPU to the volume rendering part in real time is necessary., Mar. 2008, 2008, 17, 5-8, 8
  • Not Refereed, 情報処理学会研究報告MPS-68, GPを用いた指向性フィルタの生成と評価, TAKATA Masami, Mar. 2008, 2008, 17, 21-24
  • Not Refereed, INTERNATIONAL CONFERENCE ON INFORMATICS EDUCATION AND RESEARCH FOR KNOWLEDGE-CIRCULATING SOCIETY, PROCEEDINGS, IEEE COMPUTER SOC, Application of the Kato-Temple inequality for eigenvalues of symmetric matrices to numerical algorithms with shift for singular values, Kinji Kimura; Masami Takata; Masashi Iwasaki; Yoshimasa Nakamura, The Kato-Temple inequality for eigenvalues of symmetric matrices gives a lower bound of the minimal eigenvalue lambda(m). Let A be a symmetric positive definite tridiagonal matrix defined by A = B-T B, where B is bidiagonal. Then the so-called Kato-Temple bound gives a lower- bound of the minimal singular value sigma(m) of B. In this paper we discuss how to apply the Kato-Temple inequality to shift of origin which appears in the mdLVs algorithm, for example, for computing all singular values of B. To make use of the Kato-Temple inequality a Rayleigh quotient for the matrix A = B-T B and a right endpoint of interval where lambda(m) = sigma(2)(m) belongs are necessary. Then it is shown that the execution time of mdLVs with the standard shifts can be shorten by a possible choice of the generalized Newton bound or the Kato-Temple bound., 2008, 113, +
  • Not Refereed, 情報処理学会研究報告MPS-67, 特定ユーザのための嗜好分析パターン抽出の一手法, TAKATA Masami, Dec. 2007, 2007, 128, 61-64
  • Not Refereed, IPSJ SIG technical reports, Information Processing Society of Japan (IPSJ), 複数Webページの重要文抽出および直感的理解を支援するためのGUIの開発, TAKATA Masami, Lately, Web space has been one of the most important resources for information exploration. However, users are sometimes confused because of various information floods in the Web space. In this paper, we introduce a method which extracts important sentences and keywords from plural Web documents provided from search results to acquire necessary information more effectively. Furthermore, we developed a GUI which supports visually easy and intuitive understanding for the users. Our method with the GUI reduces users' burden for web exploration drastically., Dec. 2007, 2007, 128, 81-84, 84
  • Not Refereed, 情報処理学会研究報告MPS-67, 3Dスキャナのための自動テクスチャマッピング手法, TAKATA Masami, Dec. 2007, 2007, 128, 85-88
  • Not Refereed, 情報処理学会研究報告MPS-67, リアルタイムアナグリフ動画像生成, TAKATA Masami, Dec. 2007, 2007, 128, 89-92, 92
  • Not Refereed, 情報処理学会研究報告BIO-11, E-cell systemを用いた酵母菌のグルカン合成に関するモデル構築, TAKATA Masami, Dec. 2007, 2007, 128, 113-120
  • Not Refereed, IPSJ SIG technical reports, Information Processing Society of Japan (IPSJ), 地球流体科学研究用のポータルGfdnaviへの3次元可視化機能の組み込み, TAKATA Masami, In recent years, visualization technology is essential for various scientific research. Especially, Simulation or measurement results are used as forms of visualization results to be well analyzed. In this paper, we add 3D visualization functions to Gfdnavi with outputting visualization results to client-sides through a remote X-window as well as the performance improvement with GPGPU., Dec. 2007, 2007, 128, 163-167, 166
  • Not Refereed, 平成19年度情報処理学会関西支部支部大会講演論文集, 仮想空間用入力デバイスとしてのWiiリモコンの利用, TAKATA Masami, Oct. 2007, 19-22
  • Not Refereed, 平成19年度情報処理学会関西支部支部大会講演論文集, Boid法の拡張による捕食行動の表現, TAKATA Masami, Oct. 2007, 23-26
  • Not Refereed, 平成19年度情報処理学会関西支部支部大会講演論文集, ブートストラップ法を用いたポートフォリオ分析の適用例, TAKATA Masami, Oct. 2007, 185-186
  • Not Refereed, 数理解析研究所講究録1594,非線形波動現象の数理と応用, 行列の特異値計算のmdLVsアルゴリズムにおける最近の進展, TAKATA Masami, Sep. 2007, 136-148
  • Not Refereed, 日本応用数理学会2007度年会, 特異値分解ライブラリのための最小特異値見積もり公式について, TAKATA Masami, Sep. 2007, 230-231, 231
  • Not Refereed, Proceedings of the 2007 IEICE Society Conference, ICA 適用による地電流データからの地震前駆的シグナル抽出結果に対する考察, TAKATA Masami, Sep. 2007, AS-2-4 (CD Proceeding).
  • Not Refereed, Proceedings of the 2007 IEICE Society Conference, 地震短期予測のための地電流解析Webシステムの構築, TAKATA Masami, Sep. 2007, AS-2-5 (CD Proceeding).
  • Not Refereed, IPSJ SIG Notes, Information Processing Society of Japan (IPSJ), 食事と健康状態の関連を知るための一手法, TAKATA Masami, Recently, the size of memory storage is getting larger by advancement of science. As a result, information found in reservoir data has great variety and complexity. Therefore, it is necessary to get useful information from the data, which is hard to treat by the previous statistic analyses, or the database, which has variety of formats. To do so, the data mining is useful technique. This article tries to find the relationship between ingested food and condition of health without human prejudices and make new important indicator of controlling human health by employing data mining technique. Also the experiment is held to find the association-rules between food and condition of health., Mar. 2007, 2007, 19, 77-80, 80
  • Not Refereed, IPSJ SIG Notes, Information Processing Society of Japan (IPSJ), ニンテンドーDSを用いた書字学習トレーニングソフトの開発, TAKATA Masami, We are newly developing training software for dyslexia disorder that is a type of learning disorder (LD). Person of dyslexia disorder has a trouble of the reading and writing of the character, and typical training methods are used to trace and to draw the character just on paper. The rehabilitation software for handwriting newly developed works on the portable type game machine, Nintendo DS. The repetition study on the game machine is easier than that of doing on paper. Moreover, it is an advantage that the learner can study with continued interest by adding the game. In this report, the content of the development background and the software developing now is described., Mar. 2007, 2007, 19, 81-84, 84
  • Not Refereed, 「仮想実験と大規模シミュレーションで拓く先端科学2007」高性能コンピューティング利用技術に関する総合シンポジウム予稿集, i アプリを用いた分子動力学ソフト GROMACS の構築, TAKATA Masami, Feb. 2007, 7-10
  • Not Refereed, 情報処理学会研究報告MPS-62, Auto View Selectionを実現する一手法, TAKATA Masami, Dec. 2006, 2006, 135, 49-51
  • Not Refereed, 情報処理学会研究報告MPS-62, 情報処理学会, 短期地震研究のための統合的なマルチデータベースの設計, TAKATA Masami, Dec. 2006, 2006, 135, 65-69, 68
  • Not Refereed, 情報処理学会研究報告, 熱揺らぎを考慮したドッキングシステムの開発, TAKATA Masami, Dec. 2006, 2006, 135, 119-126
  • Not Refereed, 情報処理学会研究報告, SVMを用いた金属イオン結合部位予測システムの開発, TAKATA Masami, Dec. 2006, 2006, 135, 157-164
  • Not Refereed, 日本応用数理学会2006年度年会, 対称3重対角行列の固有値分解のためのダブル分割統治法-現状と課題-, TAKATA Masami, Sep. 2006, 314-315
  • Not Refereed, 情報処理学会研究報告MPS-61, アブストラクトを用いた論文分類システムの設計と実装, TAKATA Masami, Sep. 2006, 2006, 95, 33-36
  • Not Refereed, IPSJ SIG Notes, Information Processing Society of Japan (IPSJ), 固有分解と特異値分解用ライブラリの性能評価のためのテスト行列に関する考察, TAKATA Masami, In order to evaluate the performance of numerical libraries computing eigendecomposition or singular value decomposition, we need to construct matrices of which we can compute accurate eigendecomposition or singular value decomposition. It is standard method that we compute eigendecomposition or singular value decomposition by using long precision, verify the accuracy, and compare approximate values with the accurate values. However, by using this method, we can not examine, considering purpose of use and special situation. Therefore, we introduce how to reconstruct the matrices from given solutions (eigen values/eigen vectors or singular values/singular vectors). Moreover, we introduce some special method which can be adapted to tests in larger dimension matrices., Jun. 2006, 2006, 63, 1-6, 6
  • Not Refereed, IPSJ SIG Notes, Information Processing Society of Japan (IPSJ), 新しい特異値計算ルーチンによる対称3重対角行列の固有値計算の評価, TAKATA Masami, There are some approved methods for computing eigenvalues of symmetric tridiagonal matrices. In this paper the mdLVs scheme [7] for computing singular values of bidiagonal matrices is applied to the symmetric tridiagonal eigenvalue problem and is compared with the several routines of LAPACK as a new algorithm for eigenvalues. When one needs both accuracy and computational time, the mdLVs scheme is shown to be better for the positive-definite case., Mar. 2005, 2005, 19, 115-120, 120
  • Not Refereed, 情報処理学会研究報告HPC-100, ロトカ・ボルテラ系による特異値計算アルゴリズムの並列化, TAKATA Masami, Dec. 2004, 2004, 128, 13-18
  • Not Refereed, IEICE technical report. Neurocomputing, The Institute of Electronics, Information and Communication Engineers, 誤り訂正符合のナイーブ平均場近似, TAKATA Masami, In the communication theory, solving error correcting code problem is a important subject. Recently, to reveal the charactericity of the erroro correcting code, several reseachers applied statistical-mechanical approach to this problem. In our research, we treated the error correcting code as the Bayes inference framework. We used MPM (Maximizer of the Posterior Marginals) inference, that is a kind of Bayes inference, and, moreover, applied naive mean field approximation. In the field of the artificial neural network, this approximation is applied for reduction of the computational cost by substitution of the stochastic binary units with the deteministic continuousu value units. However, there exsists few report of the performance of this approximation by quantitatively. Therefore, we evaluated the performance of this approximation by analysis and compare it with the compuater simulation., Mar. 2003, 102, 730, 143-148, 148
  • Not Refereed, 情報処理学会研究報告ARC-141-1, 乱流シミュレーションの並列化と評価, TAKATA Masami, Jan. 2001, 2001, 10, 1-6
  • Not Refereed, 情報処理学会研究報告MPS-28-6, ニューラルネットワークによるプログラム分割アルゴリズムの改善, TAKATA Masami, Feb. 2000, 2000, 16, 21-24
  • Not Refereed, 情報処理学会研究報告MPS-27-8, 実用的な近似解を与えるプログラム分割アルゴリズム, TAKATA Masami, Nov. 1999, 99, 96, 29-32
  • Not Refereed, 情報処理学会研究報告ARC-134-14, 分枝限定法を用いたプログラム分割の下界に対する考察, TAKATA Masami, Aug. 1999, 99, 67, 79-84
  • 電子情報通信学会技術研究報告 = IEICE technical report : 信学技報, 電子情報通信学会, A Stress Indexing Method by Respiratory Variability Analysis, 太田 楓; 石川 由羽; 高田 雅美, 23 Jun. 2015, 115, 112, 231, 236
  • 電子情報通信学会技術研究報告 = IEICE technical report : 信学技報, 電子情報通信学会, A Clustering Method for Gene Expression Data Based or COG Classification, 大古田 みのり; 石川 由羽; 高田 雅美, 23 Jun. 2015, 115, 112, 237, 242
  • 電子情報通信学会技術研究報告 = IEICE technical report : 信学技報, 電子情報通信学会, A Stress Indexing Method by Respiratory Variability Analysis, 太田 楓; 石川 由羽; 高田 雅美, 23 Jun. 2015, 115, 111, 115, 120
  • 電子情報通信学会技術研究報告 = IEICE technical report : 信学技報, 電子情報通信学会, A Clustering Method for Gene Expression Data Based or COG Classification, 大古田 みのり; 石川 由羽; 高田 雅美, 23 Jun. 2015, 115, 111, 121, 126
  • 映像情報メディア学会技術報告 = ITE technical report, 映像情報メディア学会, Biometrics Authentication Using the Amplitude and the Area of Heartbeat Waveform, 野村 涼子; 石川 由羽; 高田 雅美, Jun. 2015, 39, 21, 11, 16
  • 研究報告数理モデル化と問題解決(MPS), Sign Language Recognition using Leap Motion Controller, 舩阪 真生子; 石川 由羽; 髙田 雅美; 城 和貴, 本稿では,聴覚障害者が音声入力の代替手段として使用する,Leap Motion Controller を用いた指文字認識について提案する.指文字の認識に用いる条件分岐として,手指の形の特徴に着目した 19 種類を提案する.19 種類の条件分岐を用いて指文字認識アルゴリズムを作成する.条件分岐の順番によって作成されるフローチャートは異なり,条件分岐の正答率の違いから,全文字に対する平均認識率が変化する.条件分岐の並び替えは膨大な組合せであるため,指文字認識アルゴリズムの自動生成における最適解の探索に遺伝的アルゴリズムを適用する.In this paper, deaf to use as an alternative method of voice input, we propose sign language recognition using Leap Motion Controller. As decisions using sign language, we propose 19 kinds of decisions that focus on characteristic of hands and fingers. We construct sign language recognition algorithm using 19 kinds of decisions. The constructed flowchart is differing as order of decisions, recognition rate for all letters change from the difference accuracy rate of decisions. For sorting of decisions is enormous combination, we apply genetic algorithm to search for the optimal solution in the automatic construction of sign language recognition algorithm., 24 Feb. 2015, 2015, 8, 1, 6
  • 年会論文集, 日本教育情報学会, 2PC1 教材開発と知的財産権(教材開発,一般研究,教育情報と人材育成~未来を育む子供たちのために~), 吉井 直子; 篠澤 和恵; 高田 雅美; 城 和貴, 算数・数学で学習する図形(幾何分野)の理解を高めるための教材開発を行い,2013年3月に特許権を取得した.そして,開発した教材を用いて3歳児から5歳児を対象に「算数セミナー」を開催している.本稿では,教育実践と指導用テキスト開発の報告を行うとともに,知的財産権の重要性について述べる., 2015, 31, 264, 265
  • ハイパフォーマンスコンピューティングと計算科学シンポジウム論文集, 次世代HPC・Enterprise統合基盤のためのアプリ上の共通の特徴, 田邊昇; 冨森苑子; 高田雅美; 城和貴, 31 Dec. 2013, 2014, 43, 43
  • 研究報告数理モデル化と問題解決(MPS), Evaluation of Loop Transformation methods with CUDA, HATA NORIKO; TAKATA MASAMI; JOE KAZUKI, 本稿では,GPU のための最適化を施すことによる GROMACS の性能改良について述べる.その最適化には CUDA を用いる.GPU を用いた GROMACS の実行については,GROMACS-OpenMM が既に開発されている.しかしながら,ベンチマークの結果,GROMACS-OpenMM による実行した際,CPU のみで GROMACS を実行した時と比べ,ほとんど実行時間が変わらないシミュレーションや,むしろ遅くなるシミュレーションも存在する.そこで,GROMACS-OpenMM の改良を行う.GROMACS は大規模なプログラムである.そこで,本稿では,関数やメモリアクセスの最適化に焦点を当てる.In this paper, we present some performance improvement of GROMACS for GPUs, which is a well-known application for analysis of molecule, with several optimizations. The optimizations are applied for CUDA. An exsisting GROMACS for GPU, GROMACS-OpenMM has already been developed. However, by our benchmarking results, the execution time of some sample programs of GROMACS-OpenMM are nearly equal or longer than that in GROMACS on a CPU. Therefore, GROMACS with GPU should be improved. Since GROMACS consists of huge number of modules, in this paper, we focus on function calls and memory access., 05 Jul. 2010, 2010, 10, 1, 6
  • 年会論文集, 日本教育情報学会, エッジ画像における奥行き知覚を利用した教育への一提案 (教育資料・実践(2)), 吉井 直子; 高田 雅美; 城 和貴, 初等中等教育現場では,黒板を使用する機会が多い.黒板にかいたは3次元物体は,エッジ画像に近くなる.そこで,本稿ではエッジ画像を用いて奥行き知覚の能力を調べる.実験結果から回帰分析によって発達モデル式を構築する.さらに,モデル式の精度評価を行い,エッジ画像の奥行き知覚について,男子と女子の間には発達の違いがあることを理論値で示す.最後に,それを補完するために教育場面で活用できる教材を提案する., Aug. 2010, 26, 170, 173
  • 研究報告ハイパフォーマンスコンピューティング(HPC), 情報処理学会, Relaxing the Dependency of Matrix Shape on the Performance of Sparse Matrix Vector Product Using a GPU and an Extended Memory, 田邊 昇; Boonyasitpichai Nuttapon; 中條 拓伯; 小川 裕佳; 高田 雅美; 城 和貴, GPUのメモリシステムにおける不連続アクセスにおける問題を解決するために筆者らは先行研究でScatter/Gather機能を有する拡張メモリシステムを提案した。疎行列ベクトル積に対する評価も行ない,提案拡張メモリシステム側の疎行列ベクトル積に対するスループットをPCI expressのバンド幅以上にできることを示した。しかし,疎行列ベクトル積に関しては評価行列が1つのみであった。本報告ではその種類を増やした実験によって,提案方式の性能の行列形状依存性について評価した.その結果,提案方式は行列の形状に性能が鈍感で,高い実行性能を安定して提供できることがわかった.In order to overcome the problems of discontinuous accessing in memory systems of various processors, we have proposed an extended memory system which has an additional function of scattering and gathering. In this paper, we have confirmed the assumption on feasibility of throughput of the memory system is correct or not with a simulator based on DRAMsim2 with various matrices. As the result, we have confirmed proposed system has good stability for matrix figure., 08 Mar. 2011, 2011, 11, 1, 8
  • 年会論文集, 日本教育情報学会, 幼児の大画面3DCG体験学習会の報告と今後の課題 (日本教育情報学会 第27回年会 教育情報のイノベーション--デジタル世代をどう導くか) -- (初中等教育での学習教育支援・実践と情報教育), 吉井 直子; 篠沢 和恵; 高田 雅美; 城 和貴, 幼稚園教育は,新幼稚園教育要領にも記されているように環境を通して行う活動を基本としている.本稿では,我々の実施している,幼児を対象とした「大画面3DCG体験学習会」について報告する.これは,新たな環境となりつつある3DCGの最新技術に幼児が触れ,サイエンスやテクノロジーに関心を持つきっかけを与えることを目的とするものである.アンケート結果から,2D作品鑑賞と比較すると,3DCG体験の方がより幼児たちは興味や関心を持つという傾向が示唆される.また3年間実施した経験から,今後幼稚園教育で利用するための課題について述べる., Aug. 2011, 27, 150, 153
  • 年会論文集, 日本教育情報学会, 3B1 エッジ画像における幼児期の奥行き知覚の特徴(一般研究 初中等教育の情報教育1,教育情報のイノベーション~デジタル世代をどう導くか~), 吉井 直子; 高田 雅美; 城 和貴, 幼児が効果的に形の学習を行うためには,適切な教材が必要である.そこで,本稿では,物体の輪郭を取り出すエッジ抽出が施されたエッジ画像を用いて実験を行い,幼児の奥行き知覚に関する特徴を考察する.実験結果から,5歳児男子は5歳児女子より奥行きをよく知覚していること,5歳児女子は個人差が大きく,平行に並ぶものより重なりがあるものをよく知覚できるということが検証できた., 2011, 27, 258, 259
  • 研究報告数理モデル化と問題解決(MPS), The Kanji character recognition evalution for the modern book of the same pablisher, 福尾 真実; 高田 雅美; 城 和貴, 本研究では,同一出版者の近代書籍に対する漢字認識実験を行う.国立国会図書館では,近代デジタルライブラリーとして,所蔵している書籍を Web 上で公開している.これらは画像データとして公開されており,文書内容を用いた検索は行えない.そのため早急なテキスト化が求められている.書籍のスキャンは,見開きの状態で行われている.ゆえに,テキスト化を行う際,文書の傾き補正を行う必要があるが,本稿では新たな角度補正の手法を提案する.また,出版者で使用しているフォントが異なっており文字認識が困難となっている.そこで,本稿では同一出版者の近代書籍に対し漢字の認識実験を行う.In this research, we perform the kanji recognition experiment for the modern book of the same publisher. In National Diet Library, a book possessing as the digital library from meiji era is shown on Web. These are shown as image data and cannot perform the search using document contents. Therefore it is demanded an immediate text becoming it. The scan of the book is carried out in the state of the spread. In addition, fonts using in publishers are different, and character recognition becomes difficult. When we become the text of the book, it is necessary to make the degree of leaning revision of the document, but suggests technique of new angle revision in this report. Therefore we perform the recognition experiment of the kanji for the modern book of the same publisher in this paper., 12 Sep. 2012, 2012, 26, 1, 6
  • IPSJ SIG Notes, Information Processing Society of Japan (IPSJ), Estimating a Lower Bound for a Branch and Bound Based Program Partitioning Algorithm, TAKATA Masami; SAITO Tetsuya; NAKANISHI Tsuneo; JOE Kazuki, In this paper, Girkar's program partitioning algorithm for acyclic and task graphs is investigated, and a lower bound as an improvement is proposed to be validated by experimental results. In many cases, the algorithm restricted by the lower requires less execution time for search while obtaining optimal solutions. Although there were a few cases that solutions did not become optimal, the solutions were approximate because the errors to optimal solutions were small. The solutions to the task graphs which Girkar's algorithm could not be applied to for lack of memory could be obtained., 02 Aug. 1999, 1999, 67, 79, 84
  • IPSJ SIG Notes, Information Processing Society of Japan (IPSJ), A Program Partitioning Algorithm for Practical Approximate Solutions, SAITO Tetsuya; TAKATA Masami; JOE Kazuki; KUNIEDA Yoshitoshi; FUKUDA Akira, In this paper we propose an heuristics, without guarantees for the optimal solution, to obtain practical approximate solutions for the Girkar's branch-and-bound based program partitioning algorithm. Our heuristics, which focuses on a sequence of edges in a task graph, evaluates the maximum and average value of distances between edges. The values are used as an incumbent value for branching. Experimental results show our heuristics has an ability to obtain practical approximate solutions for the algorithm., 25 Nov. 1999, 1999, 96, 29, 32
  • RIMS Kokyuroku, Kyoto University, Recent Developments of the mdLVs Algorithm for Singular Values and the I-SVD Algorithm for Singular Value Decomposition (Fast Algorithms in Computational Fluids : theory and applications), Nakamura Yoshimasa; Iwasaki Masashi; Kimura Kinji; Takata Masami, Jun. 2008, 1606, 87, 104
  • IPSJ SIG Notes, Information Processing Society of Japan (IPSJ), A Method for Automatic Viewpoint Selection, NAKAGAWA MACHIKO; TAKATA MASAMI; JOE KAZUKI, Recently, three dimensional objects are widely used for visualization with various methods, as improvement of scientific techniques and visualization techniques. Since enormous and complicated data can be visualized, it is important to select a good viewpoint to get useful knowledge from the visualization results. However, there is no difinition for viewpoint selection, becouse good viewpoints are selected by the purpose and/or the target. In this paper, we improve an existing viewpoint selection model with plural properties., 21 Dec. 2006, 2006, 135, 49, 52

Works

  • 展覧会:奥行き学習教材 -1000年教材の誕生-, Jul. 2013, Jul. 2013

Awards

  • 2017年度コンピュータサイエンス領域奨励賞, 情報処理学会, 2017
  • 第110回ベストプレゼンテーション賞, 情報処理学会 数理モデルと問題解決研究会, 2016
  • プレゼンテーション賞, 情報処理学会 数理モデル化と問題解決研究会, 石川由羽;高田雅美;梅田智広;城和貴, Dec. 2011
  • プレゼンテーション賞, 情報処理学会 数理モデル化と問題解決研究会, 芳野可奈子;高田雅美;天白成一;城和貴, Mar. 2007
  • HPCS2006最優秀論文賞, IPSJ, HPC, Jan. 2006

Industrial Property Rights

  • Patent right, 特異値分解装置, TAKATA Masami, 岩崎雅史,誉田太朗,阪野真也,高田雅美,中村佳正, 特願2005-351089, Dec. 2005

Research Projects

  • 2017, 2020, 近代書籍からの知の抽出, 科学技術振興調整費による総合研究, 0, 0, 0, 基盤研究B 17H01829, Competitive research funding
  • 2014, 2016, 近代書籍自動テキスト化支援環境の構築, 科学研究費補助金, 0, 0, 0, Competitive research funding
  • 2009, 2012, 近代デジタルライブラリの自動テキスト化, 科学技術振興調整費による総合研究, 0, 0, 0, 科学研究費補助金「基盤研究C」, Competitive research funding
  • 2008, 2012, マルチコアプロセッサに対応した革新的特異値分解ライブラリの開発, 科学研究費補助金, 0, 0, 0, 基盤研究A, Competitive research funding
  • 2009, 2009, Speed-up in mdLVs by Limitation in Computational Number of Shift, 補助金, 0, 0, 0, 電気通信普及財団「海外渡航旅費援助」, Competitive research funding
  • 2009, 2009, Speed-up in mdLVs by Limitation in Computational Number of Shift, Grants and Funding, 0, 0, 0, Competitive research funding
  • 2006, 2009, 特定問題への新しい特異値分解アルゴリズムの適用に関する研究, 科学研究費補助金, 0, 0, 0, 若手研究B, Competitive research funding
  • 2001, 2001, Effective Parallelization of Turbulent Flow Simulation, 補助金, 0, 0, 0, 電気通信普及財団「海外渡航旅費援助」, Competitive research funding
  • 2001, 2001, Effective Parallelization of Turbulent Flow Simulation, Grants and Funding, 0, 0, 0, Competitive research funding
  • Grant-in-Aid for Scientific Research (B), Apr. 2020, Mar. 2023, 20H04483, Coinvestigator, 近代書籍からの知の再構築, 城 和貴; 高田 雅美; 石川 由羽, Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (B), Nara Women's University, 17680000, 13600000, 4080000, rm:published_papers;rm:published_papers;rm:published_papers;rm:presentations;rm:presentations


Copyright © MEDIA FUSION Co.,Ltd. All rights reserved.