Researchers Database
Yoshida Tetsuya
| ■Faculty | Faculty Division of Engineering Research Group of Engineering | |||
| ■Position | Professor | |||
Last Updated :2024/06/12
researchmap
Profile and Settings
Name (Japanese)
YoshidaName (Kana)
Tetsuya
Degree
Research Interests
Research Areas
Research Experience
- Apr. 2014, 9999, Nara Women's University, 教授
- Apr. 2004, Mar. 2014, Hokkaido University, Graduate School of Information Science and Technology, Division of Computer Science, 准教授
- Oct. 2001, Mar. 2004, Osaka University, The Institute of Scientific and Industrial Research, 助手
Education
- Apr. 1994, Mar. 1997, The University of Tokyo, 工学系研究科, 先端学際工学専攻, Japan
- Apr. 1991, Mar. 1994, The University of Tokyo, The Graduate School of Engineering, 航空宇宙工学専攻
- Oct. 1992, Sep. 1993, University of Edinburgh, Department of Artificial Intelligence, 知識処理専攻
- Apr. 1987, Mar. 1991, The University of Tokyo, The Faculty of Engineering, 航空学科, Japan
Teaching Experience
- Apr. 2018, 9999
- Apr. 2018, 9999
- Apr. 2017, 9999
- Apr. 2017, 9999
- Apr. 2016, 9999
- Apr. 2016, 9999
- Apr. 2015, 9999
- Apr. 2014, 9999
- Apr. 2014, Mar. 2018
- Apr. 2014, Mar. 2018
- Apr. 2014, Mar. 2016
Association Memberships
Ⅱ．研究活動実績
Published Papers
- Refereed, NICOGRAPH 2021, フルペーパー，2021, 学生奨励賞, 花紋スモッキングの展開図に対するモジュール, 夛田美彩; 吉田哲也, Nov. 2021
- Refereed, 情報処理学会論文誌：数理モデル化と応用, リジッドルームのための織物組織図の近似最適化, 吉田哲也, Aug. 2021, 14, 3, 76, 84, Scientific journal
- Refereed, 芸術科学会論文誌, 花紋スモッキングの展開図に基づく紐の生成, 吉田哲也; 藤田真奈美, Mar. 2021, 20, 1, 10, 20
- Refereed, 芸術科学会論文誌, 正多角形の貼り合わせを用いた花紋スモッキングの 組み合わせの拡張, 夛田美沙; 吉田哲也, Nov. 2020, 19, 4, 40, 48, Scientific journal
- Refereed, 芸術科学会論文誌, 花紋折りに基づくスモッキングのパターン作成と組 み合わせのデザイン, 吉田哲也,藤﨑千晶, Jun. 2020, 19, 2, 9, 24, Scientific journal
- Refereed, 日本建築学会計画系論文集, ネパー ルの世界文化遺産登録都市における都市型住居の外観意匠類型 – バクタプ ル東部の都市街区を事例に–, 飛鳥濱岡; 山本直彦; 吉田哲也; 宮内杏里; 増井正哉; 向井洋一, Jan. 2019, 84, 756, 425, 435, Scientific journal
- Refereed, COMPUTATIONAL INTELLIGENCE, WILEY, COMMUNITY STRUCTURE-BASED APPROACH FOR NETWORK IMMUNIZATION, Tetsuya Yoshida; Yuu Yamada, We propose a community structure-based approach that does not require community labels of nodes, for network immunization. Social networks have been widely used as daily communication infrastructures these days. However, fast spreading of information over networks may have downsides such as computer viruses or epidemics of diseases. Because contamination is propagated among subgraphs (communities) along links in a network, use of community structure of the network would be effective for network immunization. However, despite various research efforts, it is still difficult to identify ground-truth community labels of nodes in a network. Because communities are often interwoven through intermediate nodes, we propose to identify such nodes based on the community structure of a network without requiring community labels. By regarding the community structure in terms of nodes, we construct a vector representation of nodes based on a quality measure of communities. The distribution of the constructed vectors is used for immunizing intermediate nodes among communities, through the hybrid use of the norm and the relation in the vector representation. Experiments are conducted over both synthetic and real-world networks, and our approach is compared with other network centrality-based approaches. The results are encouraging and indicate that it is worth pursuing this path., Feb. 2017, 33, 1, 77, 98, Scientific journal, 10.1111/coin.12082
- Refereed, Computational Intelligence, A Graph-based approach for Semi-Supervised Clustering, Yoshida, T, May 2014, 30, 2, 263, 284, Scientific journal
- Refereed, International Journal of Knowledge-Based & Intelligent Engineering Systems, Rectifying the representation learned by Non-negative Matrix Factorization, Yoshida, T, Nov. 2013, 17, 4, 279, 290, Scientific journal
- Refereed, ACTIVE MEDIA TECHNOLOGY, AMT 2013, SPRINGER-VERLAG BERLIN, Toward Robust and Fast Two-Dimensional Linear Discriminant Analysis, Tetsuya Yoshida; Yuu Yamada, This paper presents an approach toward robust and fast Two-Dimensional Linear Discriminant Analysis (2DLDA). 2DLDA is an extension of Linear Discriminant Analysis (LDA) for 2-dimensional objects such as images. Linear transformation matrices are iteratively calculated based on the eigenvectors of asymmetric matrices in 2DLDA. However, repeated calculation of eigenvectors of asymmetric matrices may lead to unstable performance. We propose to use simultaneous diagonalization of scatter matrices so that eigenvectors can be stably calculated. Furthermore, for fast calculation, we propose to use approximate decomposition of a scatter matrix based on its several leading eigenvectors. Preliminary experiments are conducted to investigate the effectiveness of our approach. Results are encouraging, and indicate that our approach can achieve comparative performance with the original 2DLDA with reduced computation time., 2013, 8210, 126, 135, International conference proceedings
- Refereed, 情報処理学会論文誌:数理モデル化と応用,, 重複コミュニティ発見のための重み付き線グラフ, 吉田 哲也, Jul. 2012, 5, 3, 79, 88, Scientific journal
- Refereed, 情報処理学会論文誌:数理モデル化と応用,, ネットワークのノード情報を考慮した正則化モジュラリティ固有空 間法, 吉田 哲也, Jul. 2012, 5, 1, 65, 73, Scientific journal
- Refereed, INTELLIGENT DECISION TECHNOLOGIES (IDT'2012), VOL 2, SPRINGER-VERLAG BERLIN, Immunization of Networks via Modularity Based Node Representation, Tetsuya Yoshida; Yuu Yamada, We propose an approach for immunization of networks via modularity based node representation. Immunization of networks has often been conducted by removing nodes with large centrality so that the whole network can be fragmented into smaller subgraphs. Since contamination is propagated among subgraphs (communities) along links in a network, besides centrality, utilization of community structure seems effective for immunization. However, despite various efforts, it is still difficult to identify true community labels in a network. Toward effective immunization of networks, we propose to remove nodes between communities without identifying community labels of nodes. By exploiting the vector representation of nodes based on the modularity matrix of a network, we propose to utilize not only the norm of vectors, but also the relation among vectors. Two heuristic scoring functions are proposed based on the inner products of vector representation and their filtering in terms of vector angle. Preliminary experiments are conducted over synthetic networks and real-world networks, and compared with other centrality based immunization strategies., 2012, 16, 33, 44, International conference proceedings, 10.1007/978-3-642-29920-9_4
- Refereed, INTELLIGENT DECISION TECHNOLOGIES (IDT'2012), VOL 2, SPRINGER-VERLAG BERLIN, Line Graph for Weighted Networks toward Overlapping Community Discovery, Tetsuya Yoshida, We propose generalized line graph for weighted networks toward overlapping community discovery from the networks. Community discovery from a network has often been conducted by assigning each node in a network only to one community. However, in real world networks, a node (e.g., user) might belong to several communities. For undirected networks without self-loops, we propose to generalize line graph by defining the weights in the line graph based on the weights in the original network. Based on the line graph representation, a node can be assigned to more than one community by assigning the links adjacent to the node to the corresponding communities. Various properties of the proposed generalized line graph are clarified, and the properties indicate that our proposal is a natural extension of the conventional line graph. Preliminary experiments are conducted over several real-world networks, and the results indicate that the proposed generalized line graph can improve the quality of the discovered overlapping communities., 2012, 16, 403, 413, International conference proceedings
- Refereed, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), A comparative study of community structure based node scores for network immunization, Yuu Yamada; Tetsuya Yoshida, Network immunization has often been conducted by removing nodes with large network centrality so that the whole network can be fragmented into smaller subgraphs. Since contamination (e.g., virus) is propagated among subgraphs (communities) along links in a network, besides centrality, utilization of community structure seems effective for immunization. We have proposed community structure based node scores in terms of a vector representation of nodes in a network. In this paper we report a comparative study of our node scores over both synthetic and real-world networks. The characteristics of the node scores are clarified through the visualization of networks. Extensive experiments are conducted to compare the node scores with other centrality based immunization strategies. The results are encouraging and indicate that the node scores are promising. © 2012 Springer-Verlag., 2012, 7669, 328, 337, International conference proceedings, 10.1007/978-3-642-35236-2_33
- Refereed, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Influence of erroneous pairwise constraints in semi-supervised clustering, Tetsuya Yoshida, Side information such as pairwise constraints is useful to improve the clustering performance in general. However, constraints are not always error free in general. When erroneous constraints are specified as side information, treating them as hard constraints could have the disadvantage since strengthening incorrect or erroneous constraints can lead to performance degradation. In this paper we conduct extensive experiments to investigate the influence of erroneous pairwise constraints over various document datasets. Several state-of-the-art semi-supervised clustering methods with graph representation were evaluated with respect to the type of constraints as well as the number of constraints. Experimental results confirmed that treating pairwise constraints as hard constraints is vulnerable to erroneous ones. However, the results also revealed that the influence of erroneous constraints depends on how the constraints are exploited inside a learning algorithm. © 2012 Springer-Verlag., 2012, 7669, 43, 52, International conference proceedings, 10.1007/978-3-642-35236-2_5
- Refereed, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Community structure based node scores for network immunization, Tetsuya Yoshida; Yuu Yamada, We propose community structure based node scores for network immunization. Since epidemics (e.g, virus) are propagated among groups of nodes (communities) in a network, network immunization has often been conducted by removing nodes with large score (e.g., centrality) so that the major part of the network can be protected from the contamination. Since communities are often interwoven through intermediating nodes, we propose to identify such nodes based on the community structure of a network. By regarding the community structure in terms of nodes, we construct a vector representation of each node based on a quality measure of communities for node partitioning. Two types of node score are proposed based on the direction and the norm of the constructed node vectors. © 2012 Springer-Verlag., 2012, 7458, 899, 902, International conference proceedings, 10.1007/978-3-642-32695-0_95
- Refereed, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Overlapping community discovery via weighted line graphs of networks, Tetsuya Yoshida, We propose an approach for overlapping community discovery via weighted line graphs of networks. For undirected connected networks without self-loops, we generalize previous weighted line graphs by: 1) defining weights of a line graph based on the weights in the original network, and 2) removing self-loops in weighted line graphs, while sustaining their properties. By applying some off-the-shelf node partitioning method to the weighted line graph, a node in the original network can be assigned to more than one community based on the community labels of its adjacent links. Various properties of the proposed weighted line graphs are clarified. Furthermore, we propose a generalized quality measure for soft assignment of nodes in overlapping communities. © 2012 Springer-Verlag., 2012, 7458, 895, 898, International conference proceedings, 10.1007/978-3-642-32695-0_94
- Refereed, International Journal of Knowledge-Based and Intelligent Engineering Systems, A re-coloring approach for graph b-coloring based clustering, Tetsuya Yoshida; Hiroki Ogino, This paper proposes a re-coloring approach for graph b-coloring based clustering. Based on the notion of graph b-coloring in graph theory, a b-coloring based clustering method was proposed. However, previous method did not explicitly consider the quality of clusters, and could not find out better clusters which satisfy the properties of b-coloring. Although a greedy re-coloring algorithm was proposed to reflect the quality of clusters, it was still restrictive in terms of the explored search space due to its greedy and sequential re-coloring process. We aim at overcoming the limitations by enlarging the search space for re-coloring, while guaranteeing the b-coloring properties. In our approach, the vertices in a graph are divided into two disjoint subsets based on the properties of b-coloring. A best first re-coloring algorithm is proposed to realize non-greedy search for the admissible colors of vertices. A color exchange algorithm is proposed to remedy the problem in sequential re-coloring. These algorithms are orthogonal to each other with respect to the re-colored vertices, and thus can be utilized in conjunction. The proposed approach was evaluated against several UCI benchmark datasets. The results are encouraging and indicate the effectiveness of the proposed method, especially with respect to the ground truth micro-averaged precision. © 2012 - IOS Press and the authors. All rights reserved., 2012, 16, 2, 117, 128, Scientific journal, 10.3233/KES-2011-0236
- Refereed, 情報処理学 会論文誌:数理モデル化と応用, 特徴表現のスパース性を考慮したNMF, 木村 圭吾; 吉田 哲也, 2012, 5, 1, 21, 29, Scientific journal
- Refereed, NMFを用いた表現学習に対するコレスキー分解を用い た補正法, 吉田哲也; 荻野広樹, Jul. 2011, 4, 3, 94, 101
- Transactions of the Japanese Society for Artificial Intelligence, The Japanese Society for Artificial Intelligence, Adaptive Ripple Down Rules Method based on Description Length, YOSHIDA Tetsuya; WADA Takuya; MOTODA Hiroshi; WASHIO Takashi, A knowledge acquisition method Ripple Down Rules (RDR) can directly acquire and encode knowledge from human experts. It is an incremental acquisition method and each new piece of knowledge is added as an exception to the existing knowledge base. Past researches on RDR method assume that the problem domain is stable. This is not the case in reality, especially when an environment changes. Things change over time. This paper proposes an adaptive Ripple Down Rules method based on the Minimum Description Length Principle aiming at knowledge acquisition in a dynamically changing environment. We consider the change in the correspondence between attribute-values and class labels as a typical change in the environment. When such a change occurs, some pieces of knowledge previously acquired become worthless, and the existence of such knowledge may hinder acquisition of new knowledge. In our approach knowledge deletion is carried out as well as knowledge acquisition so that useless knowledge is properly discarded to ensure efficient knowledge acquisition while maintaining the prediction accuracy for future data. Furthermore, pruning is incorporated into the incremental knowledge acquisition in RDR to improve the prediction accuracy of the constructed knowledge base. Experiments were conducted by simulating the change in the correspondence between attribute-values and class labels using the datasets in UCI repository. The results are encouraging., 01 Nov. 2004, 19, 460, 471, 10.1527/tjsai.19.460
- Refereed, Transactions of the Japanese Society for Artificial Intelligence, Fast apriori-based graph mining algorithm and application to 3-dimensional structure analysis, Yoshio Nishimura; Takashi Washio; Tetsuya Yoshida; Hiroshi Motoda; Akihiro Inokuchi; Takashi Okada, Apriori-based Graph Mining (AGM) algorithm efficiently extracts all the subgraph patterns which frequently appear in graph structured data. The algorithm can deal with general graph structured data with multiple labels of vartices and edges, and is capable of analyzing the topological structure of graphs. In this paper, we propose a new method to analyze graph structured data for a 3-dimensional coordinate by AGM. In this method the distance between each vertex of a graph is calculated and added to the edge label so that AGM can handle 3-dimensional graph structured data. One problem in our approach is that the number of edge labels increases, which results in the increase of computational time to extract subgraph patterns. To alleviate this problem, we also propose a faster algorithm of AGM by adding an extra constraint to reduce the number of generated candidates for seeking frequent subgraphs. Chemical compounds with dopamine antagonist in MDDR database were analyzed by AGM to characterize their 3-dimensional chemical structure and correlation with physiological activity., 2003, 18, 5, 257, 268, Scientific journal, 10.1527/tjsai.18.257
- Refereed, IEEJ Transactions on Electronics, Information and Systems, A Method for Detecting Conceptual Difference based on Diverse Structure, Teruyuki Kondo; Shogo Nishida; Tetsuya Yoshida, When people carry out collaborative works, conceptual difference due to different backgrounds and knowledge can hinder their communication and deteriorate collaboration. We have been carrying out research on detecting conceptual difference by focusing on the situation in which different symbols are used to denote the same meaning and the same symbols are used to denote different meaning. In our approach each user's knowledge is represented as a decision tree respectively and difference in concept is detected as difference in the structure of decision trees. This paper points out some problems in constructing a unique decision tree based on a single information criterion. Based on the idea of diverse structure, this paper proposes a new method for increasing the performance of detection by constructing multiple decision trees with diverse structure. Genetic algorithm is utilized to realize the idea of diverse structure and experiments with motor diagnosis cases on the implemented system confirmed the i provement on the performance of detection. © 2003, The Institute of Electrical Engineers of Japan. All rights reserved., 2003, 123, 2, 345, 354, Scientific journal, 10.1541/ieejeiss.123.345
- Not Refereed, The transactions of the Institute of Electronics, Information and Communication Engineers. D-I, The Institute of Electronics, Information and Communication Engineers, A Method for Detecting Conceptual Difference Based on Correlation between Decision Trees, OHNISHI Kensuke; YOSHIDA Tetsuya; NISHIDA Shogo, 複数の人間が共同作業を行う場合には,背景知識や専門領域の違いから派生する概念の相違が意思疎通を阻害する要因となる.筆者らは従来から概念の相違として異なるシンボルを同じ意味に用いたり,同じシンボルを異なる意味に用いたりする場合を考え,ユーザの知識を決定木として表現し,概念の相違を決定木の構造の相違としてとらえることでその検出を行うことに取り組んできた.本論文では,従来の各ユーザの知識から仮想的な知識を生成して決定木を構築することにより相違を検出する手法における問題点を指摘し,その解決策として決定木におけるクラス分類の行われ方からボトムアップに決定木間の相関関係を構築して相違検出を行う手法を提案する.提案した手法をプロトタイプに実装し,モータの故障診断事例に対する実験を通じて検出精度の向上を確認した., 25 Jul. 2002, 85, 8, 784, 797
- Refereed, 電子情報通信学会論文誌D-I, 決定木の相関関係に基づいた概念相違検出手法, 2002, vol.J85, No.8, pp.784-797, Scientific journal
- Refereed, Human interface. The Transaction of Human Interface Society, ヒュ-マンインタフェ-ス学会, An Information Acquisition Support System based on Keyword Relationship, SHINKAI Daiki; YOSHIDA Tetsuya; NISHIDA Shogo, 2002, Vol.4, No.4,pp.19-30, 4, 207, 217, Scientific journal
- Refereed, Human interface. The Transaction of Human Interface Society, ヒュ-マンインタフェ-ス学会, An Image Based Support System for Web Page Design, WATANABE Masato; YOSHIDA Tetsuya; SAIWAKI Naoki; HIJIKATA Yoshinori; NISHIDA Shogo, 2001, Vol.3, No.4,pp.287-297, 4, 73, 83, Scientific journal
- Refereed, ADVANCES IN WEB-AGE INFORMATION MANAGEMENT, PROCEEDINGS, SPRINGER-VERLAG BERLIN, Utilizing the correlation between query keywords for information retrieval, T Yoshida; D Shinkai; S Nishida, This paper proposes a method to utilize the correlation between query keywords in information retrieval toward extrapolating their semantic information. In accordance with the rapid development of internet and WWW (World Wide Web), it has been getting more and more hard to pinpoint the appropriate document from the huge information resource. Various search engines have been developed to retrieve the appropriate information based on the keywords, however, it is hard for the user to specify the keywords enough to pinpoint the appropriate documents. Since there often exist some semantic correlation between the keywords, this paper proposes to finding out the another keyword by utilizing their semantic correlation in order to narrow the scope of seach. Experiments were carried out to investigate the effectiveness of the proposed method and the result hinted the effectiveness of our approach as a pre-processing to narrow the scope of search for search engines., 2001, 2118, 49, 59, Scientific journal
- Not Refereed, The Transactions of the Institute of Electrical Engineers of Japan. C, 電気学会, Adaptive Hypermedia System for Supporting Information Providers to Direct Users through Hyperspace, HIJIKATA Yoshinori; YOSHIDA Tetsuya; NISHIDA Shogo, 01 Nov. 2000, 120, 11, 1720, 1731
- Refereed, ヒューマンインタフェース学会論文誌, ユーザ参加型設計のそれの事例を利用した意図理解支援インタフェース, 2000, 2, 2, 87, 95, Scientific journal
- Refereed, IEEE RO-MAN 2000: 9TH IEEE INTERNATIONAL WORKSHOP ON ROBOT AND HUMAN INTERACTIVE COMMUNICATION, PROCEEDINGS, IEEE, A method for supporting software design based on comment on software, T Yoshida; K Hashimoto; T Yamaoka; S Nishida, This paper proposes a method for supporting software design by utilizing the comment which is usually put on software. It is believed that the software designer often leaves the hint or clue with respect to the reason or rationale for each module or component as the comment on software when he/she designs and implements the software. The comment is utilized for structuring: software into a tree and CBR (Case-Based Reasoning) is carried out based on the tree structure for enabling; the effective re-use of past software with respect to the structure and organization of software in the uppser stream in design. A prototype system has been implemented with the proposed support method and experiments have been carried out to investigate the effectiveness of our approach., 2000, 311, 315, International conference proceedings
- Refereed, The Transactions of the Institute of Electronics, Information and Communication Engineers. A, 一般社団法人電子情報通信学会, A Communication Model in Emergency Which Considers Competence, Duty, Responsibility and Knowledge, KOISO Takashi; YOSHIDA Tetsuya; SAIWAKI Naoki; NISHIDA Shogo, 大規模災害に対処する防災システムを考えるとき, 適切な部署に,適切な情報が適切なタイミングで届くコミュニケーシヨンの実現が重要である. 本論文では, 階層型組織での緊急時の意思決定について分析するとともに, 組織の各構成員のもつ権限・義務・責任・知識に注目することにより, 人的構造を考慮した緊急時のコミュニケーションモデルを提案する. 更にこのコミュニケーションモデルのプラント制御への適用例を示すとともに, このモデルの適用可能性として緊急時のコミユニケーシヨン支援システムへの応用や, 組織形態のコミュニケーシヨンの視点から見た評価についても論じる., Mar. 1999, 82, 3, 445, 453, Scientific journal
- Refereed, NEW DIRECTIONS IN ROUGH SETS, DATA MINING, AND GRANULAR-SOFT COMPUTING, SPRINGER-VERLAG BERLIN, Evolving granules for classification for discovering difference in the usage of words, T Yoshida; T Kondo; S Nishida, This paper proposes an evolutionary approach for discovering difference in the usage of words to facilitate collaboration among people. When people try to communicate their concepts with words, the difference in the meaning and usage of words can lead to misunderstanding in communication, which can hinder their collaboration. In our approach each granule of knowledge in classification from users is structured into a decision tree so that difference in the usage of word can be discovered as the difference in the structure of tree. By treating each granule of classification knowledge as an individual in Genetic Algorithm (GA), evolution is carried out with respect to the classification efficiency of each individual and diversity as a population so that difference in the usage of words will emerge as the difference in the structure of decision tree. Experiments were carried out on motor diagnosis cases with artificially encoded difference in the usage of words and the result shows the effectiveness of our evolutionary approach., 1999, 1711, 366, 374, Scientific journal
- Refereed, IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences, Learning the Balance between Exploration and Exploitation via Reward, 1999, 82-EA, 11, 2538, 2545, Scientific journal
- Refereed, IPSJ Journal, 一般社団法人情報処理学会, Design of the Interfaces to Detect Conceptual Difference among Different People(
on Next Generation Human Interface and Interaction), KONDO TERUYUKI; YOSHIDA TETSUYA; NISHIDA SHOGO, Conceptual difference among different people is a serious problem when people work in collabolation with others. In this paper, we study about conceptual difference among different people and describe a method of detecting conceptual difference by the decision trees constructed by users' knowledge in the cases that the same symbols are used in the different meaning and the cases that the different symbols are used in the same meaning. And a prototype sysytem is evaluated., May 1998, 39, 5, 1195, 1202, Scientific journal - Refereed, IEICE TRANSACTIONS ON FUNDAMENTALS OF ELECTRONICS COMMUNICATIONS AND COMPUTER SCIENCES, IEICE-INST ELECTRONICS INFORMATION COMMUNICATIONS ENG, A cooperation method via metaphor of explanation, T Yoshida; K Hori; S Nakasuka, This paper proposes a new method to improve cooperation in concurrent systems within the framework of Multi-Agent Systems (MAS). Since subsystems work concurrently, achieving appropriate cooperation among them is important to improve the effectiveness of the overall system. When subsystems are modeled as agents, it is easy to explicitly deal with the interactions among them since they can be modeled naturally as communication among agents with intended information. Contrary to previous approaches which provided the syntax of communication protocols without semantics, we focus on the semantics of cooperation in MAS and aim at allowing agents to exploit the communicated information for cooperation. This is attempted by utilizing more coarse-grained communication based on the different perspective for the balance between Formality and richness of communication contents so that each piece of communication contents can convey more meaningful information in application domains. In our approach agents cooperate each other by giving Feedbacks based on the metaphor of explanation which is widely used in human interactions, in contrast to previous approaches which use direct orders given by the leader based on the pre-defined cooperation strategies. Agents show the difference between the proposal and counterproposals for it, which are constructed with respect to the Former and given as the feedbacks in the easily understandable terms For the receiver. From the comparison of proposals agents retrieve the information on which parts are agreed and disagreed by the relevant agents, and reflect the analysis in their following behavior. Furthermore, communication contents are annotated by agents to indicate the degree of importance in decision making for them, which contributes to making explanations or feedbacks more understandable. Our cooperation method was examined through experiments on the design of micro satellites and the result showed that it was effective to some extent to facilitate cooperation among agents., Apr. 1998, E81A, 4, 576, 585, Scientific journal
- Refereed, 情報処理学会論文誌, 複数の人間における概念相違検出のためのインターフェースの設計, 近藤, 1998, 39, 5, 1195, 1202, Scientific journal
- Refereed, INTERNATIONAL CONFERENCE ON MULTI-AGENT SYSTEMS, PROCEEDINGS, IEEE COMPUTER SOC, A reinforcement learning approach to cooperative problem solving, T Yoshida; K Hori; S Nakasuka, We propose an extension of reinforcement learning methods to cooperative problem solving in multi-agent systems. Exploiting multiple agents for complex problems is promising, however, learning is necessary since complete domain knowledge is rarely available. The temporal difference algorithm is applied in each agent to learn a heuristic evaluation of states. Besides the reward for solutions produced by agents, we define the reward for coherence as a. whole and exploit them to facilitate cooperation among agents for global problem solving. We evaluate the method by experiments on the satellite design problem. The result shows that our method enables agents to learn to cooperate as well as to learn individual heuristics within one framework. Especially, agents themselves learn to take the appropriate balance between exploration and exploitation in problem solving, which is known to greatly affect the performance. It also suggests the possibility of controlling the global behavior of multi-agent systems via rewards in reinforcement learning., 1998, 479, 480, International conference proceedings
- Refereed, ARTIFICIAL INTELLIGENCE IN REAL-TIME CONTROL 1997, PERGAMON PRESS LTD, Interactive interfaces to detect conceptual difference for group knowledge acquisition, S Nishida; T Yoshida; T Kondo, Conceptual difference is a serious problem in group knowledge acquisition systems, especially when different people with different background participate in a group. This paper deals with conceptual difference and proposes a method to detect it in the cases that different symbols are used as the same meaning and/or same symbols are used as the different meanings. In section 2, conceptual difference is defined and system architecture for detection is described. In section 3, detecting algorithm is designed, and then a prototype system and its evaluation are discussed in section 4. Copyright (C) 1998 IFAC., 1998, 177, 181, International conference proceedings
- Refereed, Journal of Architecture and Planning (Transactions of AIJ), Architectural Institute of Japan, EXTERIOR DESIGN OF MAIN HOUSES WITH REGARD TO BUILDING LAYOUT IN RURAL HOUSE COMPOUNDS IN SPECIAL PRESERVATION AREAS FOR HISTORIC LANDSCAPE: The case of six village sections such as Okuyama, Asuka, Kawahara, Noguchi, Oka and Shimasho in Asuka village, YAMAMOTO Naohiko; HIRAO Kazuhiro; YOSHIDA Tetsuya; MUROSAKI Chie, Asuka village in Nara prefecture is well-known for its richness in historic remains originated from Asuka period. This paper aims to clarify the exterior design composition of rural main houses in six village sections in Asuka village. Based on the data collected in these village sections, Items/Categories list is compiled. Using this IC list data, as a result of multiple correspondence analysis followed by cluster analysis, samples are classified into 15 clusters. The analysis next searches the relation between house compound types and exterior design patterns of main houses. Finally, the paper tried to figure out each village section’s characteristic., Jul. 2022, 87, 797, 1271, 1281, Scientific journal, 10.3130/aija.87.1271
- Refereed, Journal of Fiber Bioengineering and Informatics, Generation of Approximate Weave Diagrams via Warp Pick Up Assignment, Tetsuya Yoshida, Jul. 2022, 15, 2, 79, 90, Scientific journal
- Refereed, 芸術科学会論文誌, ねじり折りに対するモジュールに基づく花紋スモッキングの組合せ, 吉田哲也,夛田 美沙, Jun. 2022, 21, 2, 77, 86, Scientific journal
- Refereed, 情報処理学会論文誌: 数理モデル化と応用, 綜絖枠数の制約下での織物組織図の近似, 吉田哲也, Jul. 2022, 15, 3, 90, 96, Scientific journal
- Refereed, 情報処理学会論文誌:数理モデル化と応用, 呼吸の特徴量を用いた心拍間隔の欠損補完, 野村涼子; 吉田哲也, Jul. 2022, 15, 3, 11, 18, Scientific journal
- Refereed, IEEE RO-MAN 2000: 9TH IEEE INTERNATIONAL WORKSHOP ON ROBOT AND HUMAN INTERACTIVE COMMUNICATION, PROCEEDINGS, IEEE, A method for supporting web page design based on impression of web page, M Watanabe; T Yoshida; N Saiwaki; S Nishida, This paper proposes an approach for web page design support by focusing: on the impression of web page with respect to the selection of color and font. Previous approaches for web page design support often focus on the automation of HTML coding: and the pasting of graphics and icons. However, the selection of color and font also greatly affect the impression of web page even with the same (information) content. The implemented system with the proposed support method tries to acquire the taste or preference of the user from the web pages which are selected by him/her during net surfing; and utilize it for web page design with the prefered impression., 2000, 13, 17, International conference proceedings
- Refereed, SMC '97 CONFERENCE PROCEEDINGS - 1997 IEEE INTERNATIONAL CONFERENCE ON SYSTEMS, MAN, AND CYBERNETICS, VOLS 1-5, I E E E, Detection algorithm for conceptual difference among different persons - as an example of emergence in the human interaction, T Yoshida; T Kondo; N Saiwaki; S Nishida, We consider emergence in the human interaction by taking the discovery of conceptual difference as an example. Conceptual difference is a serious problem in group knowledge acquisition systems, especially when different people with different background participate in a group. We deal with the conceptual difference, in which different symbols are used to express the same meaning and/or the same symbol is used to express different meanings. We propose a method to detect such types of conceptual difference by utilizing decision trees which are constructed with ID3 algorithm and GA (Genetic Algorithm). Our method is discussed in terms of the realization of emergence in the human interaction. We tackle the ''vague'' concept of emergence in the human interaction by constructing ''concrete'' structures to express conceptual difference. Our method ensures the variety in the conceptual structures by utilizing GA to explore various possibilities. The various structures constructed by the system play the role of pointing out conceptual difference from another viewpoint, which is different from those of the members in the group. This function enables the members to understand each other as well as their own concepts, which is expected to contribute to hitting new ideas as a group and to promoting the activity of the group., 1997, 1216, 1221, International conference proceedings
- Refereed, DESIGN OF COMPUTING SYSTEMS: SOCIAL AND ERGONOMIC CONSIDERATIONS, ELSEVIER SCIENCE BV, Interactive interfaces to detect conceptual difference for knowledge acquisition, T Kondo; N Saiwaki; T Yoshida; S Nishida, Conceptual difference is a serious problem in group knowledge acquisition systems, especially when different people with different background participate in a group. In our former research, each a peace of knowledge held by people is expressed by a single decision tree, and conceptual difference is detected by checking the difference. Sometimes it occurs that the attributes and the values with conceptual difference do not appear in decision trees and fails to detect conceptual difference. In this paper, we propose a method to solve this problem by using decision trees with diverse structures which are produced by our genetic algorithm., 1997, 21, 303, 306, International conference proceedings
- Advances in Knowledge Discovery and Data Mining(PAKDD), Springer, Classifier Construction by Graph-Based Induction for Graph-Structured Data., Warodom Geamsakul; Takashi Matsuda; Tetsuya Yoshida; Hiroshi Motoda; Takashi Washio, 2003, 52, 62, International conference proceedings, 10.1007/3-540-36175-8_6
- PRICAI 2002: Trends in Artificial Intelligence(PRICAI), Springer, Knowledge Discovery from Structured Data by Beam-Wise Graph-Based Induction., Takashi Matsuda; Hiroshi Motoda; Tetsuya Yoshida; Takashi Washio, 2002, 255, 264, International conference proceedings, 10.1007/3-540-45683-X_29
- PRICAI 2002: Trends in Artificial Intelligence(PRICAI), Springer, Extension of the RDR Method That Can Adapt to Environmental Changes and Acquire Knowledge from Both Experts and Data., Takuya Wada; Tetsuya Yoshida; Hiroshi Motoda; Takashi Washio, 2002, 218, 227, International conference proceedings, 10.1007/3-540-45683-X_25
- Refereed, Journal of Fiber Bioengineering and Informatics, Petaloid Folding of General Triangles for Combinatorial Smocking Design, Yoshida, T, Sep. 2022, 15, 3, 145, 156
- Refereed, Textile Research Journal, SAGE Publications, A method for approximating weave diagrams under heald frame constraint, Tetsuya Yoshida, Woven fabric is produced by interlacing the warp yarn and the weft yarn, and their relationship on woven fabric is represented as a weave diagram. In order to produce woven fabric with looms, the necessary number of heald frames can be calculated based on the weave diagram for the woven fabric. However, when the number of available heald frames is limited, it is difficult to produce complex woven fabric. To alleviate this problem, this article proposes a method for approximating weave diagrams under heald frame constraint in terms of the alternating optimization in machine learning. By defining the objective function based on the matrix representation of weave diagram, an alternating optimization method is proposed to generate an approximate weave diagram under the specified number of heald frames. The proposed method is implemented with R language, and evaluated in terms of the performance of the proposed algorithm and woven fabric. Experimental results indicate that the proposed method is effective in generating approximate weave diagrams so that they can be fabricated using looms with a limited number of heald frames., 20 Jul. 2023, 93, 15-16, 3527, 3537, Scientific journal, 10.1177/00405175221149684
- Refereed, 日本建築学会計画系論文集, 修景に向けた民家の外観意匠類型化手法と景観評価, 鈴木 裕子; 山本 直彦; 吉田哲也, Oct. 2023, 88, 812, 2751, 2761
- Refereed, Japan Architectural Review, Typology on exterior design of townhouses in a World Cultural Heritage Site of a Nepali City, Asuka Hamaoka; Naohiko Yamamoto; Tetsuya Yoshida; Anri Miyauchi; Masaya Masui; Yoichi Mukai, Oct. 2023, 6
- Refereed, 芸術科学会論文誌, 模様の回転と帯の長さに基づく花紋スモッキングの動的可視化, 吉田 哲也; 中西彩子, Nov. 2023, 22, 4, 10:1, 10:11
MISC
- Not Refereed, Journal of Japanese Society for Artificial Intelligence, The Japanese Society for Artificial Intelligence, Active Mining for Structured Data(
Active Mining), Motoda Hiroshi; Ho Tu Bao; Washio Takashi; Yada Katsutoshi; Yoshida Tetsuya; Ohara Kouzou, 01 Mar. 2005, 20, 2, 172, 179 - Not Refereed, IPSJ SIG Notes. ICS, Information Processing Society of Japan (IPSJ), Extracting Diagnostic Knowledge from Hepatitis Data by Decision Tree Graph-Based Induction, GEAMSAKUL Warodom; YOSHIDA Tetsuya; OHARA Kouzou; MOTODA Hiroshi; WASHIO Takashi, Decision Tree Graph-Based Induction (DT-GBI) is a technique for constructing a decision tree from graph-structured data. In DT-GBI, substructures (discriminative patterns) are extracted by stepwise pair expansion (pair-wise chunking) and used as test attributes at nodes of a decision tree. We applied DT-GBI to a classification task of hepatitis data. In the first experiment, the stages of fibrosis are used as classes and a decision tree is constructed for discriminating patients with F4 (cirrhosis) from patients with the other stages using only the time sequence data of blood inspection. In..., 14 Sep. 2003, 2003, 90, 53, 58
- Not Refereed, 知識ベ-スシステム研究会, 人工知能学会, Extracting Diagnostic Knowledge from Hepatitis Data by Decision Tree Graph-Based Induction (小特集 「アクティブマイニング」および一般), Geamsakul Warodom; 吉田 哲也; 大原 剛三, 14 Sep. 2003, 61, 0, 53, 58
- Not Refereed, IEICE technical report. Artificial intelligence and knowledge-based processing, The Institute of Electronics, Information and Communication Engineers, Extracting Diagnostic Knowledge from Hepatitis Data by Decision Tree Graph-Based Induction, GEAMSAKUL Warodom; YOSHIDA Tetsuya; OHARA Kouzou; MOTODA Hiroshi; WASHIO Takashi, Decision Tree Graph-Based Induction (DT-GBI) is a technique for constructing a decision tree from graph-structured data. In DT-GBI, substructures (discriminative patterns) are extracted by stepwise pair expansion (pair-wise chunking) and used as test attributes at nodes of a decision tree. We applied DT-GBI to a classification task of hepatitis data. In the first experiment, the stages of fibrosis are used as classes and a decision tree is constructed for discriminating patients with F4 (cirrhosis) from patients with the other stages using only the time sequence data of blood inspection. In..., 07 Sep. 2003, 103, 304, 53, 58
- Not Refereed, Journal of Japanese Society for Artificial Intelligence, The Japanese Society for Artificial Intelligence, MLnet, 吉田 哲也, 01 Jul. 2003, 18, 4, 470, 470
- Not Refereed, IPSJ SIG Notes. ICS, Information Processing Society of Japan (IPSJ), Functional Extension of Decision Tree - Graph-Based Induction, GEAMSAKUL Warodom; MATSUDA Takashi; YOSHIDA Tetsuya; MOTODA Hiroshi; WASHIO Takashi, A machine learning technique called Graph-Based Induction (GBI) efficiently extracts typical patterns from graph-structured data by stepwise pair expansion (pairwaise chunking). Meanwhile, a decision tree is an effective means of data classification from which rules that are easy to understand can be obtained. However, a decision tree could not be produced for the data which is not explicitly expressed with attribute-value pairs. In this paper, we propose a method of constructing a classifier (decision tree) for graph-structured data by GBI. In our approach attributes, namely substructures ..., 13 Mar. 2003, 2003, 30, 93, 98
- Not Refereed, IPSJ SIG Notes. ICS, Information Processing Society of Japan (IPSJ), Correlation Analysis of 3-dimensional Chemical Structure and its Activity by AGM, NISHIMURA Yoshio; WASHIO Takashi; YOSHIDA Tetsuya; MOTODA Hiroshi; INOKUCHI Akihiro; OKADA Takashi, Apriori-based Graph Mining (AGM) algorithm efficiently extracts all the subgraph patterns which frequently appear in graph structured data. The algorithms can deal with general graph structured data with multiple labels of vartices and edges, and is capable of analyzing the connective structure of graphs. We have proposed a faster algorithm of AGM by adding an extra constraint to reduce the number of generated candidates for seeking frequent subgraphs. In this paper, we propose a new method to analyze graph structured data which are represented with a 3-dimensional coordinate by AGM. In thi..., 13 Mar. 2003, 2003, 30, 99, 103
- Not Refereed, 知識ベ-スシステム研究会, 人工知能学会, Decision Tree-Graph-Based Inductionの機能拡張 (知識ベースシステム研究会(第60回) 人工知能基礎論研究会(第52回) 小特集:「データマイニング」および一般) -- (文部科学省科学研究費特定領域研究 情報洪水時代におけるアクティブマイニングの実現), Geamsakul Warodom; 松田 喬; 吉田 哲也, 13 Mar. 2003, 60, 0, 93, 98
- Not Refereed, 知識ベ-スシステム研究会, 人工知能学会, AGMによる3次元構造と生理活性の相関解析 (知識ベースシステム研究会(第60回) 人工知能基礎論研究会(第52回) 小特集:「データマイニング」および一般) -- (文部科学省科学研究費特定領域研究 情報洪水時代におけるアクティブマイニングの実現), 西村 芳男; 鷲尾 隆; 吉田 哲也, 13 Mar. 2003, 60, 0, 99, 103
- Not Refereed, IEICE technical report. Artificial intelligence and knowledge-based processing, The Institute of Electronics, Information and Communication Engineers, Functional Extension of Decision Tree-Graph-Based Induction, GEAMSAKUL Warodom; MATSUDA Takashi; YOSHIDA Tetsuya; MOTODA Hiroshi; WASHIO Takashi, A machine learning technique called Graph-Based Induction (GBI) efficiently extracts typical patterns from graph-structured data by stepwise pair expansion (pairwaise chunking). Meanwhile, a decision tree is an effective means of data classification from which rules that are easy to understand can be obtained. However, a decision tree could not be produced for the data which is not explicitly expressed with attribute-value pairs. In this paper, we propose a method of constructing a classifier (decision tree) for graph-structured data by GBI. In our approach attributes, namely substructures ..., 07 Mar. 2003, 102, 710, 41, 46
- Not Refereed, IEICE technical report. Artificial intelligence and knowledge-based processing, The Institute of Electronics, Information and Communication Engineers, Correlation Analysis of 3-dimensional Chemical Structure and its Activity by AGM, NISHIMURA Yoshio; WASHIO Takashi; YOSHIDA Tetsuya; MOTODA Hiroshi; INOKUCHI Akihiro; OKADA Takashi, Apriori-based Graph Mining (AGM) algorithm efficiently extracts all the subgraph patterns which frequently appear in graph structured data. The algorithms can deal with general graph structured data with multiple labels of vartices and edges, and is capable of analyzing the connective structure of graphs. We have proposed a faster algorithm of AGM by adding an extra constraint to reduce the number of generated candidates for seeking frequent subgraphs. In this paper, we propose a new method to analyze graph structured data which are represented with a 3-dimensional coordinate by AGM. In thi..., 07 Mar. 2003, 102, 710, 47, 51
- Not Refereed, マテリアルインテグレ-ション, ティ-・アイ・シ-, グラフ構造データからのマイニング (特集 大阪大学産業科学研究所 マテリアルインテグレーション--材料・生体・情報の融合を目指して(2)), 元田 浩; 鷲尾 隆; 吉田 哲也, Aug. 2002, 15, 8, 59, 62
- Not Refereed, マテリアルインテグレ-ション, ティ-・アイ・シ-, 専門家とデータの両方からの統一的知識獲得 (特集 大阪大学産業科学研究所 マテリアルインテグレーション--材料・生体・情報の融合を目指して(2)), 吉田 哲也; 元田 浩; 鷲尾 隆, Aug. 2002, 15, 8, 63, 66
- Not Refereed, IPSJ SIG Notes. ICS, Information Processing Society of Japan (IPSJ), A Faster Apriori-based Graph Algorithm, Nishimura Yoshio; Washio Takashi; Yoshida Tetsuya; Motoda Hiroshi; Inokuchi Akihiro, Apriori-based Graph Mining (AGM) algorithm derives suhgraph patterns efficiency which frequently appear in database consisting of graph structure data. In this paper, we propose a new and faster algorithm of ACM achieved by adding a condition to generate candidate frequent graphs. Simulation experiments were carried out to evaluate the performance of the proposed new ACM algorithm over the synthesized data and real world chemical data. Efficient reduction of the computation time by the proposed method has been confirmed., 23 May 2002, 2002, 45, 11, 16
- Not Refereed, IPSJ SIG Notes. ICS, Information Processing Society of Japan (IPSJ), Knowledge Discovery from Hepatitis Data based on Graph Structure, Matsuda Takashi; Yoshida Tetsuya; Motoda Hiroshi; Washio Takashi, Table representation is not suitable to represent data with many missing values in knowledge discovery, since the missing values are explicitly represented in the table and they can hinder the appropriate mining process. Graph structure can he robust for dealing with data with many missing values since they can just be omitted from the graph and thus are not explicitly represented. This paper reports a preliminary attempt to discover useful time sequence pattern from hepatitis data, which have many missing values, with Graph-Based Induction (GBI) method. Although some useful patterns for cl..., 23 May 2002, 2002, 45, 67, 72
- Not Refereed, 知識ベ-スシステム研究会, 人工知能学会, Apriori-based Graph Mining アルゴリズムの高速化 (テーマ:「アクティブマイニング」および一般), 西村 芳男; 鷲尾 隆; 吉田 哲也, 23 May 2002, 56, 0, 11, 16
- Not Refereed, 知識ベ-スシステム研究会, 人工知能学会, グラフ構造に着目した肝炎データからの知識発見 (テーマ:「アクティブマイニング」および一般), 松田 喬; 吉田 哲也; 元田 浩, 23 May 2002, 56, 0, 67, 72
- Not Refereed, Human interface. The Transaction of Human Interface Society, ヒュ-マンインタフェ-ス学会, Supporting Mutual Understanding in Participatoty Design Using Cases, YAMAOKA Takayuki; TSUJINO Katsuhiko; YOSHIDA Tetsuya; NISHIDA Shogo, 01 May 2000, 2, 2, 87, 95
- Not Refereed, Adaptive Hypermedia System for Supporting Information Providers to Direct Users through Hyperspace, 2000, 120-C, 11, 1720, 1731
- Not Refereed, Supporting Mutual Understanding in Participatory Design Using Cases, 2000, 2, 2, 87, 95
- Not Refereed, IEICE TRANSACTIONS ON FUNDAMENTALS OF ELECTRONICS COMMUNICATIONS AND COMPUTER SCIENCES, IEICE-INST ELECTRONICS INFORMATION COMMUNICATIONS ENG, Learning the balance between exploration and exploitation via reward, T Yoshida; K Hori; S Nakasuka, This paper proposes a new method to improve cooperation in concurrent systems within the framework of Multi-Agent Systems (MAS) by utilizing reinforcement learning. When subsystems work independently and concurrently, achieving appropriate cooperation among them is important to improve the effectiveness of the overall system. Treating subsystems as agents makes it easy to explicitly deal with the interactions among them since they can be modeled naturally as communication among agents with intended information. In our approach agents try to learn the appropriate balance between exploration and exploitation via reward, which is important in distributed and concurrent problem solving in general. By focusing on how to give reward in reinforcement learning, not the learning equation, two kinds of reward are defined in the context of cooperation between agents, in contrast to reinforcement learning within the framework of single agent. In our approach reward for insistence by individual agent contributes to facilitating exploration and reward for concessi:on to other agents contributes to facilitating exploitation. Our cooperation method was examined through experiments on the design of micro satellites and the result showed that it was effective to some extent to facilitate cooperation among agents by letting agents themselves learn the appropriate balance between insistence and concession. The result also suggested the possibility of utilizing the relative magnitude of these rewards as a new control parameter in MAS to control the overall behavior of MAS., Nov. 1999, E82A, 11, 2538, 2545
- Not Refereed, 知識ベ-スシステム研究会, 人工知能学会, 決定木間のリンクを利用した概念相違発見手法 (テーマ:「インターネットとAI」および一般), 大西 健介; 吉田 哲也; 西田 正吾, Mar. 1999, 43, 69, 74
- Not Refereed, 知識ベ-スシステム研究会, 人工知能学会, 検索キーワードの補完情報を利用した情報獲得支援 (テーマ:「インターネットとAI」および一般), 新開 大樹; 吉田 哲也; 西田 正吾, Mar. 1999, 43, 105, 110
- Not Refereed, The Third Pacific-Asia Conference on Knowledye Discovery and Data Mining(PAKDD-99), Discovering Conceptual Defferences among Different People via Diverse Structures, YOSHIDA T., 1999, 494, 498
- Not Refereed, The Transactions of the Institute of Electronics,Information and Communication Engineers. A, The Institute of Electronics, Information and Communication Engineers, A Communication Model in Emergency which Considers Competence, Duty, Responsiblity and Knowledge, KOISO Takashi; YOSHIDA Tetsuya; SAIWAKI Naoki; NISHIDA Shogo, 大規模災害に対処する防災システムを考えるとき, 適切な部署に,適切な情報が適切なタイミングで届くコミュニケーシヨンの実現が重要である. 本論文では, 階層型組織での緊急時の意思決定について分析するとともに, 組織の各構成員のもつ権限・義務・責任・知識に注目することにより, 人的構造を考慮した緊急時のコミュニケーションモデルを提案する. 更にこのコミュニケーションモデルのプラント制御への適用例を示すとともに, このモデルの適用可能性として緊急時のコミユニケーシヨン支援システムへの応用や, 組織形態のコミュニケーシヨンの視点から見た評価についても論じる., 1999, 82A, 3, 445, 453
- Not Refereed, 全国大会講演論文集, Information Processing Society of Japan (IPSJ), A Dynamic Linkage Method for Hypermedia Based on Metadata and User Model, Hijikata Yoshinori, 05 Oct. 1998, 57, 3, 105, 106
- Not Refereed, IEICE TRANSACTIONS ON FUNDAMENTALS OF ELECTRONICS COMMUNICATIONS AND COMPUTER SCIENCES, IEICE-INST ELECTRONICS INFORMATION COMMUNICATIONS ENG, A cooperation method via metaphor of explanation, T Yoshida; K Hori; S Nakasuka, This paper proposes a new method to improve cooperation in concurrent systems within the framework of Multi-Agent Systems (MAS). Since subsystems work concurrently, achieving appropriate cooperation among them is important to improve the effectiveness of the overall system. When subsystems are modeled as agents, it is easy to explicitly deal with the interactions among them since they can be modeled naturally as communication among agents with intended information. Contrary to previous approaches which provided the syntax of communication protocols without semantics, we focus on the semantics of cooperation in MAS and aim at allowing agents to exploit the communicated information for cooperation. This is attempted by utilizing more coarse-grained communication based on the different perspective for the balance between Formality and richness of communication contents so that each piece of communication contents can convey more meaningful information in application domains. In our approach agents cooperate each other by giving Feedbacks based on the metaphor of explanation which is widely used in human interactions, in contrast to previous approaches which use direct orders given by the leader based on the pre-defined cooperation strategies. Agents show the difference between the proposal and counterproposals for it, which are constructed with respect to the Former and given as the feedbacks in the easily understandable terms For the receiver. From the comparison of proposals agents retrieve the information on which parts are agreed and disagreed by the relevant agents, and reflect the analysis in their following behavior. Furthermore, communication contents are annotated by agents to indicate the degree of importance in decision making for them, which contributes to making explanations or feedbacks more understandable. Our cooperation method was examined through experiments on the design of micro satellites and the result showed that it was effective to some extent to facilitate cooperation among agents., Apr. 1998, E81A, 4, 576, 585
- Not Refereed, Software Design Support by Utilizing Comment, 1998, 113, 118
- Not Refereed, 7th IFAC SYMPOSIUM on Analysis, Design and Evaluation of Man-Machine Systems, Facilitate Cooperation of Humans and Machines from the viewpoint of Multi-Agent Systems, 1998, 485, 490
- Not Refereed, DISCOVERY SCIENCE, SPRINGER-VERLAG BERLIN, Discovering conceptual differences among people from cases, T Yoshida; T Kondo, We propose a method for discovering conceptual differences (CD) among people from cases. In general different people seem to have different ways of conception and thus can have different concepts even on the same thing. Removing CD seems especially important when people with different backgrounds and knowledge carry out collaborative works as a group; otherwise they cannot communicate ideas and establish mutual understanding even on the same thing. In our approach knowledge of users is structured into decision trees so that differences in concepts can be discovered as the differences in the structure of trees. Based on the candidates suggested by the system with our discovering algorithms, the users then discuss each other on differences in their concepts and modify them to reduce the differences. CD is gradually removed by repeating the interaction between the system and users. Experiments were carried out on the cases for motor diagnosis with artificially encoded CD. Admittedly our approach is simple, however, the result shows that our approach is effective to some extent as the first step toward dealing with the issue of CD among people., 1998, 1532, 162, 173
- Refereed, 7th IEEE International Workshop on Robot and Human Communication(ROMAN'98), An Extension of Reinforcement Learning in Multi-Agent Systems, 1998, 221, 226, Summary international conference
- Not Refereed, INTELLIGENT AUTONOMOUS SYSTEMS, I O S PRESS, A cooperation method to exploit design rationales for agents, T Yoshida; K Hori; S Nakasuka, This paper proposes a new method to improve cooperation in Multi-Agent Systems (MAS) from the viewpoint of Design Rationale (DR). In our approach agents cooperate each other by giving feedbacks on the proposal from each agent indirectly, in contrast to previous approaches which use direct orders given by the leader based on the pre-defined cooperation strategies. Indirect cooperation is realized by constructing and giving counter-proposals from relevant agents on the proposal. Feedbacks given as the counter-proposals play the role of pointing out insufficient portions in the original one from different perspectives toward the agreement among agents. Based on the comparison of proposals agents retrieve the information on which parts are agreed and disagreed by the relevant agents, and reflect the analysis in their following behavior. Furthermore, communication contents are annotated by agents to indicate the degree of importance in decision making for them, which contributes to making feedbacks more understandable. Our cooperation method was examined through experiments on the design of micro satellites and the result showed that it was effective to some extent to facilitate cooperation among agents., 1998, 647, 654
- Not Refereed, 3RD ASIA PACIFIC COMPUTER HUMAN INTERACTION, PROCEEDINGS, IEEE COMPUTER SOC, Supporting mutual understanding in collaborative design project, T Yamaoka; K Tsujino; T Yoshida; S Nishida, In collaborative design, participants usually have different backgrounds and standpoints. Due to the differences, it is hard for the participants mutually to understand the intention and they have to make much effort to reach a mutual agreement on the design, then the efficiency of design processes is decreased. In this paper, we propose a framework to support the participants for making the mutual understanding by grasping and showing differences among their intention in collaboration. A system based on the proposed framework has the characteristics: facilitating mutual understanding among participants by presuming their intention and by communicating them with the understandable form, visualizing the difference in concepts among participants, and making progress collaboration in an interactive way., 1998, 132, 137
- Not Refereed, Design of the Interfaces to Detect Conceptual Differnce among Defferent People, 1998, 39, 5, 1195, 1202
- Not Refereed, Technical report of IEICE. HCS, The Institute of Electronics, Information and Communication Engineers, Evaluation of organizational structure in emergency from the viewpoint of communication, Koiso Takashi; Saiwaki Naoki; Yoshida Tetsuya; Nishida Shogo, It is very important to support communications in the emergency situation for large scale systems. We focus on the evaluation of the organizational structure for emergent situations from the viewpoint of communication. Our approach emphasize quantitative analysis which contracts to the qualitative analysis in social science. First, we explain communication model briefly, and discuss how to evaluate organizations by using the model. Furthermore some prototype system are explained., 13 Jun. 1997, 97, 99, 5, 12
- Not Refereed, 11th European Conference on Artificial Intelligence, Coloured rippling : an extension of a theorem proving heuristics, 1994, 85, 89
- 研究報告数理モデル化と問題解決（MPS）, Topic Graph based Transfer Learning via generalized KL divergence based NMF, 木村 圭吾; 吉田 哲也, 本稿では，トピックグラフに基づく転移学習法を拡張し，一般化 KL (Kullback-Leibler) ダイバージェンスに基づく NMF (Non-negative Matrix Factorization) を用いた転移学習法を提案する．ダイバージェンスを通じた転移学習の確率的な解釈を目指して，フロベニウスノルムに基いてトピックの関係 （トピックグラフ） を活用する転移学習法を拡張し，転移学習を一般化KLダイバージェンスに基づく最適化問題として定式化する．最適化規準に対する補助関数を定義し，補助関数から最適化アルゴリズムを導出し，その収束性を示す．提案法を文書クラスタリングに適用し，他手法との比較を通じて提案法の有効性を示す．特に，提案法による転移学習を通じてダイバージェンスを用いた場合でも精度向上を実現できることを示す．We propose a topic graph based transfer learning method based on Non-negative Matrix Factorization (NMF) with generalized Kullback-Leibler (KL) divergence. In this paper we extend the previous NMF based transfer learning method by utilizing generalized KL divergence based NMF so that better probabilistic interpretation can be obtained with the divergence. The proposed method is formalized as the minimization of an objective function under the divergence, and an auxiliary function for the objective function is defined. From the auxiliary function, we derive a learning algorithm with multiplicative update rules, which are guaranteed to converge. The proposed method is evaluated in terms of document clustering over several well-known benchmark datasets. Especially, one drawback of generalized KL divergence based NMF algorithms is performance degradation compared with Frobenius based ones. The experimental results show that, by utilizing the topic graph, the proposed method enables to boost up the performance even with KL divergence based NMF through transfer learning., 08 Sep. 2011, 2011, 3, 1, 6
- 研究報告バイオ情報学（BIO）, A Graph Model for mutual information based clustering and its evaluation, YOSHIDA TETSUYA, 本稿では，相互情報量に基づくクラスタリング問題に対するグラフモデルを提案する．相互情報量から導出される定常分布に着想を得たデータ間の類似度関数を定義してデータ集合を辺重み付きグラフとして表現することにより，データが一様分布する場合にはハードクラスタリング問題が提案するグラフモデルにおける組合せ最適化問題に近似できることを示す．提案するグラフモデルを文書クラスタリングでのベンチマークデータである 20 Newsgroup のデータに対して評価し，他手法との比較を通じて提案手法の妥当性と有効性を確認した．We propose a graph model for data clustering based on mutual information. Based on the stationary distribution induced from the problem setting, we propose a similarity function among data objects, and represent the entire objects as an edge-weighted graph. We show that, in hard assignment, the problem can be approximated as a combinatorial problem over the proposed graph when data is uniformly distributed. The proposed approach is evaluated on the text clustering problem over the 20 Newsgroup benchmark data. The results are encouraging and indicate the effectiveness of our approach., 10 Dec. 2009, 2009, 30, 1, 7
- 人工知能学会全国大会論文集, 人工知能学会, Feature Costruction for Classification Learing from Structured Data by Graph-Based Induction, 松田 喬; 元田 浩; 吉田 哲也, 2002, 16, 1, 4
- 人工知能学会全国大会論文集, 人工知能学会, Case Generation Method for Constructing an RDR Knowledge Base and it's Evaluation, 藤原 啓成; 吉田 哲也; 元田 浩, 2002, 16, 1, 4
- 人工知能学会全国大会論文集, 人工知能学会, Experiments on Ripple Down Rules Method that Adapts to Changes in Class Distribution, 和田 卓也; 吉田 哲也; 元田 浩, 2002, 16, 1, 4
- 研究報告数理モデル化と問題解決（MPS）, Non-negative Matrix Factorization with Sparse Features, 木村 圭吾; 吉田 哲也, 本稿では，特徴表現のスパース制約を考慮した NMF (Non-negative Matrix Factorization) を提案する．近年，要素が非負である実行列を，同じく要素が非負である実行列の積として表現する非負値行列分解 (NMF) が注目を集めている．従来の研究では NMF における非負性制約が非零の要素が少ないスパースな特徴表現の学習に寄与すると考えられ，またスパース制約を導入した手法も提案されているが，これまで特徴表現のスパース性は明示的には考慮されてこなかった．本稿では NMF における特徴表現に着目し，特徴表現のスパース性を独立性と相関から定式化し，定式化したスパース性を正則化項として活用する手法を提案する．提案法を文書クラスタリングに適用し，従来法との比較を通じて提案法の有効性を示す．We propose an approach for Non-negative Matrix Factorization (NMF) with sparseness constraints on features. It has been believed that the non-negativity constraint in NMF contributes to making the learned features sparse. In addition, several approaches incorporated additional sparseness constraints, by hoping that the constraints make the features more sparse. However, previous approaches have mostly focused on coefficients, and have not considered the sparsity of features explicitly. Our approach explicitly incorporates the sparsity of features, in terms of independence of features and correlation of features. The proposed notion of sparsity is formalized as regularization terms in the framework of NMF, and learning algorithms with multiplicative update rules are proposed. The proposed approach is evaluated in terms of document clustering over well-known benchmark datasets. Several experiments have been conducted on the datasets, and comparison with other state-of-the-art NMF algorithms is reported. The results are encouraging and show that the proposed approach improves the clustering performance, while sustaining relatively good quality of data approximation., 08 Sep. 2011, 2011, 2, 1, 6
- 情報科学技術レターズ, Forum on Information Technology, LD-006 Incremental Neighborhood Graph Construction based on the Localized Update, Hacid Hakim; Yoshida Tetsuya, 22 Aug. 2007, 6, 103, 106
- 人工知能学会全国大会論文集, 人工知能学会, Knowledge Acquisition from Graph Structured Data with Constraints, 茂木 明; 吉田 哲也; ジアムサクン ワロドム, 2004, 18, 1, 4
- 人工知能学会全国大会論文集, 人工知能学会, Improvement of Search Capability of Decision Tree-Graph-Based Induction, ジアムサクン ワロドム; 松田 喬; 吉田 哲也, 2003, 17, 1, 4
- 知識ベ-スシステム研究会, 人工知能学会, Decision Tree Graph-based Induction法による肝炎データからの診断知識発見 (特集 「医療及び化学情報マイニング」および一般), Geamsakul Warodom; 吉田 哲也; 大原 剛三, 01 Mar. 2004, 64, 47, 54
- ヒュ-マンインタフェ-スデザイン研究会, 人工知能学会, 視点の異なる参加者による協同設計の支援手法の提案, 山岡 孝行; 辻野 克彦; 吉田 哲也, Jun. 1997, 31, 19, 24
Presentations
- Oral presentation
- Oral presentation
- Invited oral presentation, 24 Mar. 2023, 26 Mar. 2023
- Nominated symposium
- 吉田哲也, 情報処理学会 第132回数理モデル化と問題解決研究発表会, リジッドルームのための織物組織図の近似学習, Oral presentation, 01 Mar. 2021, 01 Mar. 2021, 02 Mar. 2021
- 夛田美沙; 吉田哲也, NICOGRAPH2020, 正多角形の貼り合わせを用いた花紋スモッキングの組み合わせの拡張, Oral presentation, 03 Nov. 2020, 01 Nov. 2020, 03 Nov. 2020
- 吉田哲也, 情報処理学会研究会数理モデル化と問題解決（MPS）, 綜絖枠数の制約下での織物組織図の近似, Oral presentation, 03 Mar. 2022, 03 Mar. 2022, 04 Mar. 2022
- 野村涼子; 吉田哲也, 情報処理学会研究会数理モデル化と問題解決（MPS）, 呼吸特徴量を用いた心拍間隔の欠損補完, Oral presentation, 13 Dec. 2021, 13 Dec. 2021, 13 Dec. 2021
- Oral presentation, 24 Mar. 2023, 26 Mar. 2023
- 南 柚衣; 橋本 朋子; 吉田哲也, 日本機械繊維学会 第75回年次大会, ATR-FTIRスペクトルの主成分分析によるセルロース繊維の接触冷感性と分子構造との関係性評価, 02 Jun. 2022, 02 Jun. 2022, 03 Jun. 2022
Awards
Research Projects
- 2021, 2023, Principal investigator
- 2021, 2023, 21K12542, Principal investigator
- 2015, 2017, 15K00307, Principal investigator
- 2018, 2020, 18K11436, Principal investigator
- 2018, 2020, 18K11436, Principal investigator
- 2018, 2020, 18K11436, Principal investigator
- 2015, 2017, 15K00307, Principal investigator
- Ripple Down Rules法による知識獲得, 0, 0, 0, Competitive research funding
- データベースからの知識発見, 0, 0, 0, Competitive research funding
- Knowledge Discovery from Databases, 0, 0, 0, Competitive research funding
- Learning in Multi-Agent Systems, 0, 0, 0, Competitive research funding
- 2021, 2023, 21K12542, Principal investigator
- 基盤研究(C), 01 Apr. 2018, 31 Mar. 2023, 18K11436, 線グラフに基づくネットワークからの多重コミュニティ発見, 吉田 哲也, 日本学術振興会, 科学研究費助成事業, 奈良女子大学, 4290000, 3300000, 990000, 利用可能なデータの量や種類の増加に効率的に対処するため，多様な価値観に応じてネットワーク上の資源を効果的に活用するための技術の確立が求められている．そこで，このような技術の確立を目指して，本研究では，線グラフに基づくネットワークからの多重コミュニティ発見の定式化，定式化に基づく多重コミュニティ発見アルゴリズムの開発，開発するアルゴリズムの計算機システムとしての実装，の実現に取り組む． 上記の目的を実現するために，本年度は，申請者が従来から研究を進めてきたグラフ構造に基づく学習手法を発展させることを通じて，グラフ構造に基づいて多重コミュニティを発見するアルゴリズムの開発と実装に取り組んだ．具体的には，下記の項目を実施した． （１） データとして与えられた重み付きグラフに対して，グラフ上の酔歩によって生成される自己ループを削除し，削除した重みの分配に基づいて定義する重み付き線グラフを構築するプログラムを計算機上にプロトタイプシステムとして実装していたが，その実装の改良に取り組んだ． （２） コミュニティ発見における標準的な評価指標の拡張として，前年度までに実装した確率的なラベル割り当てに基づいた多重コミュニティ発見指標（多重モジュラリティと呼）を計算するプログラムを改良した． （３） 社会ネットワーク分析で一般に用いられるモジュラリティを一般化して定義した多重モジュラリティに対する定式化した最適化問題に対し，最適化問題を解くアルゴリズムを前年度までに計算機上にプロトタイプシステムとして実装したが，その実装の改良に取り組んだ．, kaken
- Grant-in-Aid for Scientific Research (B), 01 Apr. 2015, 31 Mar. 2019, 15H05225, Fundamental Study on Conservation on Townhouses and Urban Environment in the World Heritage Cities in Nepal, YAMAMOTO Naohiko; MIYAUCHI Anri; PANT Mohan Moorti; SUWAL Ram Prasad, Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, Nara Women's University, 15860000, 12200000, 3660000, The aim of the research is to clarify development process of urban blocks and their townhouses from the field survey in the old city of Bhaktapur in Nepal registered as World Cultural Heritage. The areal extent of distribution of townhouses as units forming an urban block is judged by the votary area of Hindu gods and occupational surnames. Based on measurement survey of townhouses within this extent, historic development of the urban block is analyzed with regard to the historic difference of townhouses’ floor planning. This also leads to the findings that townhouses have different facades between opposite sides of a street/lane. The historic urban landscape of Nepali cities is formed in such a way that closely relates with the development process of urban blocks., kaken
- Grant-in-Aid for Scientific Research (C), 01 Apr. 2015, 31 Mar. 2018, 15K00307, Graph-based Information Theoretic Semi-Supervised Learning, Yoshida Tetsuya; IMAI Hideyuki, Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, Nara Women's University, 4680000, 3600000, 1080000, In order to cope with increasing quantity and variety of data, it is important to develop information technology which enables effective use of domain knowledge. We have developed a graph-based information theoretic semi-supervised learning method. In the developed method, the relationship among data is represented as a graph based on mutual information, and domain knowledge is regarded as constraints and used for regularization. Under the framework of optimization learning, we have developed a semi-supervised learning algorith based on the representation matrix of the graph. The algorithm has been implemented as a prototype system, and experiments over the prototype system were conducted over several benchmark datasets. The results indicate the effectiveness of the developed learning method., kaken
- Grant-in-Aid for Scientific Research (A), 2006, 2007, 19200007, Flexible aggregate extraction and aggregate federation of knowledgeresources over the Web and the integrated generic technologies for their reuse andvisualization, TANAKA Yuzuru; LUNZER ARAN EDWARD; ITO Kimihito; YOSHIDA Testuya, Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, Hokkaido University, 47320000, 36400000, 10920000, 群抽出・群連携の理論を確立し、これに基づくジェネリックな技術を開発し、さらには、これらの技術を用いたウェブ上の知的資源の統合再利用・可視化基盤技術を確立した。具体的には, (a)ウェブ上の知的資源の群抽出・群連携理論を確立し、(b)ウェブ上の知的資源の群抽出・群連携インタラクティブシステムを開発し、(c)ウェブ上の知的資源の統合再利用基盤技術を確立し、(d)ウェブ上の知的資源の統合可視化基盤技術を確立し、(e)統合再利用・可視化基盤技術の応用と評価を行った。, kaken
- 若手研究(B), 2006, 2007, 18700131, 最小記述長原理に基づきデータからの帰納学習を統合する知識獲得技術の研究, 吉田 哲也, 日本学術振興会, 科学研究費助成事業, 北海道大学, 3300000, 3300000, 本年度に行った研究実績を以下に述べる. 1.昨年度に実施した検討に基づいて,知識の整合性を保障しながら,人問の専門家から継続的に知識を獲得できる知識獲得手法に対して,機械学習の分野における主要技術の一つである事例(データ)からの学習に着目して,人間の専門家からの知識獲得(これも広義の意味での学習とみなせる)と,データからの学習の両者とを統合化するプログラムコードの実装を更に進めた. 2.1.での実装においては,蓄積した知識の根拠となる事例集合をも保持するような知識べースの表現形式を用いることによって,知識べースを表現する木構造における各ノードを,事例に対する推論(判断)を行う知識とともに,その根拠となる事例集合をも保持することを実現した. 3.各ノードが保持するそれぞれの知識の複雑さを,知識表現の複雑さに対する記述長と,その知識を用いて与えられた事例集合を分類した際の精度(予測誤差)に対する記述長として,2.で実現した各ノードに蓄積する事例集合を利用して計算するモジュールをコードとして実装した. 4.2.,3.,などを通じて実装したモジュールを利用することにより,人間の専門家とデータからの両者から知識を獲得できる知識獲得手法を,計算機上にプロトタイプシステムとして実現した. 5.上記を通じて開発したプロトタイプシステムを,機械学習の分野で公開されているベンチマークデータなどに対して適用し,その性質や挙動を評価した., kaken
- Grant-in-Aid for Scientific Research (A), 2005, 2006, 17200004, Federal Integration and Advanced Reusing Technologies for Scientific and Technological Knowledge Resources including Retrieval and Computation Services, TANAKA Yuzuru; LUNZER Aran; ITOH Kimihito; YOSHIDA Tetsuya, Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, Hokkaido University, 45110000, 34700000, 10410000, Before this research project, there were no enabling technologies for the federal integration and reuse of knowledge resources to extract related knowledge fragments from knowledge resources over the Web, and to relate them for their federal integration into a new reusable knowledge resource. Users needed to manually make a copy of each output of a service and to put it into an input form of another service. Or otherwise, they needed to rely on Web Service technologies. In the latter case, API of the servers needs to be publicized and some programming effort is required, which may work for the definition of routine knowledge integration, but cannot work for ad hoc knowledge federation. Some services may graphically out put their results. There are increasing needs for the knowledge federation of services with 2D and 3D GUIs. This research aimed at the establishment of the generic technologies for the federal integration and advanced reuse of scientific and technological knowledge resources including the retrieval and computation services such as information retrieval, database, and simulation systems. In order to achieve this goal, we have developed the following 5 generic software technologies: (1) interactive extraction of related knowledge fragments over the Web into a tabular form, (2) federation and integration of extracted knowledge fragments, (3) federation and integration of extracted knowledge resources with legacy applications, (4) meta information description about the federation and integration capabilities, and automatic federation based on such descriptions, and (5) publication and distribution infrastructure for newly composed knowledge resources. The Web has mutually related knowledge resources embedded in Web pages as a list in a text or as inputs and outputs of the same web application. This project established a new technology to extract them as interoperable and mutually relatable resources. We proposed a new representation format named the knowledge media view to represent extracted knowledge resources and to federate them. For legacy applications with 2D and 3D GUIs, we have developed a new view integration technologies by extending the shadow copy technologies from those for the integration of 2D and 2D applications to those for the 2D and 3D applications integration and for the 3D and 3D application integration. We have related these technologies to the knowledge media view representation, and established the knowledge federation technologies between an arbitrary knowledge fragments extracted from the Web and an arbitrary legacy application., kaken
- Grant-in-Aid for Scientific Research (B), 2004, 2005, 16300046, Development of Constructive Induction Method of Useful Attributes from Complex Structured Data, MOTODA Hiroshi; WASHIO Takashi; YOSHIDA Tetsuya; OHARA Kouzou, Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, Osaka University, 13800000, 13800000, In data mining where a set of useful knowledge is to be mined from a huge amount of data, the standard practice is to use the original attribute which is used in the original data representation. However, it often happens that the original attributes are not expressive enough and constructing new attributes from the original ones is inevitable. This is called feature construction and yet a better method is to be found. In this research a new feature construction method that is interleaved in the construction of a decision tree is developed and its performance is tested using both artificial and real world datasets. Since the forms of the data to handle become diversified and graph is a good way to represent data of general form, a graph mining method based on sequential chunking method is coupled with a decision tree construction method. The subgraph found at each decision node can be considered as a constructed attribute. The biggest problem of being unable to find overlapping patterns by the straightforward chunking can be avoided by devising pseudo-chunking. The resulting CI-GBI (Chunkingless Graph-based Induction) is now able to do complete search by setting the values for the parameters appropriately. Since it does not use the notion of anti-monotonicity of subgraph subsumption, it can find subgraphs which other state-of-the-art approaches cannot find. Further, because it is guaranteed that the frequency counting of the found subgraphs is accurate, various indices that use frequency, e.g. information gain, are also evaluated accurately and CI-GBI becomes better suited as a feature construction component in decision tree construction. Subgraph search is called recursively during the tree construction and the best feature is constructed on the fly at each decision node. Compared with the straightforward chunking approach, the size of the constructed tree becomes much smaller and the predictive accuracy for an unseen instance becomes better. The application to the chronic hepatitis dataset indicated that it is indeed possible to predict the liver cirrhosis by blood test alone., kaken
- 若手研究(B), 2002, 2004, 14780280, ラベル付き・ラベル無しデータを統合的活用する共学習システムの構築方法に関する研究, 吉田 哲也, 日本学術振興会, 科学研究費助成事業, 3800000, 3800000, 研究実績は以下のとおり. 1.初年度,次年度に検討した,グラフ構造データを対象とする学習アルゴリズム(ラベル付きデータに対するDecision Tree Graph-based Induction (DT-GBI)法,および,ラベル無しデータに対するBeam-wise Graph-based Induction (B-GBI法))実問題に適用した場合,抽出される知識(部分グラフ)は例外を多く含み,正確性に欠けるという問題があった.この問題を解決するため,学習アルゴリズムで抽出した部分グラフを制約として更に活用し,得られた知識(部分グラフ)の精緻化を行う方法を考案し,実装した. 2.1.を実現する際,制約として活用する部分グラフの包含関係を確認するために部分グラフの同型問題を解く必要があるが,この問題はNP完全であり,大規模なデータを対象とする場合に計算時間が問題となる.この問題に対し,GBI法はもともとGreedy探索に基づく手法あることを考慮し,正確性よりも高速性を重視し,GBI法を活用した部分グラフ同型問題の近似解法を考案し,実装し,1.の手法に導入した. 3.次年度に引き続き,実問題への適用例として,千葉大学医学部付属病院から提供いただいた肝炎データに適用した.各患者の肝臓の線維化程度,肝炎の型(B・C)に加え,新たにインターフェロンの効用予測問題に適用した.DT-GBI法により構築した決定木の分類精度を評価し,また,DTGBI法およびB-GBI法で抽出した部分グラフを1.の手法を用いて精緻化した知識(部分グラフ)の妥当性を,専門家(医師)により評価いただいた. 4.本研究課題を通じて開発した手法および得られた知見(UCIのpromoterデータと3.の肝炎データを対象)を整理して海外の学術論文に投稿し,受理された., kaken
- 萌芽研究, 2002, 2004, 14658102, 多様な形式データからの特徴抽出に基づく一元的検索手法の開発, 鷲尾 隆; 元田 浩; 吉田 哲也, 日本学術振興会, 科学研究費助成事業, 大阪大学, 3400000, 3400000, 本年度の研究実績は以下の通りである。 1.データ形式を超えた検索手法の開発 前年度までに,画像情報など二次元配列データに関して,二次元のビット配列形式を解いて単なるビット配列に変換し,共通のデータフォーマットを有する特徴量に変換する手法を開発したが,最終年度は画像に限らず,テキスト文書を含む一般の非暗号化バイナリーデータに関して,データから形式依存のビット配列情報を捨象し,残された情報を数学的な不変量に縮約して特徴量に変換する手法を確立した。ビット配列情報から規則順序形式を捨象し一般的なビット配列に変換した.更に数学的不変量を抽出し,検索の手がかりとなる特徴ベクトルを構成した.また,最終年度はデータ形式を超えた高速検索を可能にするべく,被検索データのデータ構造と検索アルゴリズムの開発を行った.特徴ベクトルから高速に情報検索することができるように,いずれの特徴ベクトルがいずれのデータから得られたものであるかを紐付けする逆引きファイルを構成した.そして,検索時には実データを見ることなく逆引きファイル情報を参照することで,高速な検索を可能とした。これにより,種々の構造を有するデータ形式に適用可能な高速検索手法を得た. 2.検索システムのプロトタイプ作成による性能評価と手法修正 上記で新たに開発した手法やアルゴリズムをデータサーバ計算機にプログラムとして実装した.性能評価として検索精度及び速度を評価した.その結果,前年度には二次元配列データなどの構造データに関しては数分単位の検索時間が必要とされるたが,最終年度は上記の手法開発により大幅な高速化が図られ,数秒で構造データの検索が可能になった.更に二次元配列構造に限らず,テキストや系列構造,木構造,グラフ構造など,多様な構造データに関して検索性能を検証し,いずれに関しても所与の性質,類似性を持った構造データを高速に検索できることを確認した. 以上により,本研究の当初の目的である既存のデータ形式に留まらず将来新たに生み出されるであろうデータ形式にも対応しうる,データ内容に共通した不変な数学的特徴を抽出する原理,それによって類似性を判定する原理,及びそれらに基づく検索手法が得られた., kaken
- 特定領域研究, 2001, 2004, 13131206, 構造データからのアクティブマイニング, 元田 浩; 鷲尾 隆; 大原 剛三; TUBAO Ho; 矢田 勝俊; 吉田 哲也, 日本学術振興会, 科学研究費助成事業, 大阪大学, 64800000, 64800000, 研究実績は以下のとおり. 1.グラフ構造データからの決定木構築プログラムDT-GBIでの探索過程に領域知識の制約(指定したパターンを含む,含まない)を導入した. 2.ペアを仮想ノードとして扱い,チャンキングをせず探索する新グラフマイニング手法Cl-GBIを開発した.適切なパラメータ設定により完全探索が可能になり,GBIの数え落としの問題点などを解決した. 3.上記Cl-GBIを組み込んだ決定木構築プログラムDT-ClGBIを開発し,肝炎データセットで性能を評価した. 4.数値データを伴うデータから,数値を記号離散化することなしに相関の高い数値区間を自動抽出する原理を確立し,それに基づく数値相関規則導出手法を開発した. 5.ユーザ指向データマイニングシステムD2MSの肝炎患者に関するルールの理解容易性向上を確認し,多数のルールから統計的に有意なものを選定する手法とルール学習において領域知識を表現の制約に加える手法を提案した, 6.科学データマイニングとしてゲノムおよび結晶データを並行して解析した.前者に関しては,SVMによるタンパク質の2次構造におけるβターンの予測手法を拡張しγターンを予測した.後者に関しては,粉末回折データから結晶構造を同定する手法を遺伝的アルゴリズムに基づき開発した. 7.意味的まとまりを捉えたパッセージの集合として文書を表し,トレランス・ラフ集合モデルによるソフトマッチを導入し,意味を反映した相関ルールを得る手法を開発した. 8.グラフマイニング手法AGMを消費者行動データに適用し,アクティブマイニングによる実証実験を行い新しいデータを収集した.アルコール市場分析から得られた知見に基づき,実際の店舗で店頭プロモーションを行った結果,対象商品の売り上げ増加,関連商品の同時購入頻度の増加を検証することができた., kaken
- Grant-in-Aid for Scientific Research (B), 2001, 2003, 13558034, Development of Knowledge Acquisition System that can Adapt to Environment Change, MOTODA Hiroshi; SATOH Ken; YOSHIDA Tetsuya; WASHIO Takashi; TERABE Masahiro, Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, Osaka University, 13600000, 13600000, In this study an attempt is made to integrate a knowledge acquisition technique which is based on the notion of refinement of existing knowledge introducing the finding from cognitive science and an inductive learning technique which has been developed in the field of machine learning to induce a classifier from accumulated data, to propose a new knowledge acquisition technique to fuse these two different knowledge sources into an operational knowledge, and to verify its effectiveness using real world datasets. More concretely, the following study has been conducted: l) to study a method in which there is no need to know how the knowledge has been acquired and stored in the knowledge base and it is assured that the acquisition of new knowledge does not cause the problem of inconsistency with the existing knowledge, 2) to study a method to conduct continuous knowledge acquisition while automatically identifying which pieces of knowledge have become useless and deleting them still maintaining the overall consistency and the understandability of the constructed knowledge base, 3) to study a method to utilize the accumulated data in such a way that switching between two different knowledge sources (i.e. human exert and accumulated data) can be made at any time of knowledge acquisition without rebuilding the knowledge base from scratch and adapt to environment changes. The developed system has been tested against many datasets of different properties and confirmed to exhibit satisfactory performance. It is now possible to start constructing a knowledge base system acquiring initial pieces of knowledge from human expert and then switching to inductive learning later when abundant data have been accumulated. System developer no more need worry about which pieces of knowledge to delete., kaken
- Grant-in-Aid for Scientific Research (B), 2000, 2002, 12480097, Research on Spatio-tenporal Data Management Method for Moving objects, NISHIDA Shogo; YOSHIDA Tetsuya; SAIWAKI Naoki, Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, Osaka University, 5300000, 5300000, The management of spatio-temporal data is becoming increasingly important in many application fields such as Computer-Aided Design (CAD), Geographic Information Systems(GlS) and Virtual Reality(VR) systems, because capturing data on moving objects has become easier by the development of GPS and PDA. The head investigator and his colleagues have been studying on the efficient spatio-temporal data management method based on tree structure. In this research, we have developed a new data management system for moving objects and investigated its application. The contents of our research include the following items. 1) Spatio-temporal data management structure 2) Data management system for camera 3) Information filtering, kaken
- Grant-in-Aid for Scientific Research (B), 2000, 2002, 12480088, DEVELOPMENT OF A GRAPH STRUCTURE DATAMINING METHOD AND IDENTIFICATION SYSTEM OF ACTIVE MOLECULE SUBSTRU CTURES, WASHIO Takashi; YOSHIDA Tetsuya; OKADA Takashi; MOTODA Hiroshi, Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, OSAKA UNIVERSITY, 15400000, 15400000, In the first fiscal year, the theoretical framework of graph structure data mining was investigated, and a prototype system for active molecule substructure identification was developed. In this work, the representation of graph structure data in computers and search principle of characteristic graph patterns are studied. Them, the survey of techniques in chemistry which can be introduced to our work has been conducted, and these techniques were reflected in the prototype system. Finally, the basic performance of the prototype system has been evaluated through the substructure extraction in carcinogenetic and mutagenetic chemical component data. In the nest fiscal year, the framework of the graph structure data mining was extended to be more efficient in terms of computation time and memory consumption, and the real scale system for active molecule substructure identification has been developed. The algorithm for the efficient computation time and memory consumption was developed, and under the comparison with the conventional techniques in chemistry, the function of the real system was designed. Then, the principle and the algorithm of the real system was modified and extended to enable the graph structure data mining on the massive graph structure data. In the final fiscal year, further functions desired to be implemented in the view of chemical analysis were investigated based on the real system developed in the former year, and some functions which can be implemented feasibly were added to the real system. Then, from the view points of the chemical engineering and the computational theory, the practicality and the wide applicability of the real system have been evaluated. Through these evaluations, the practical and high performance of the developed real sysem has been confirmed. The effort to develop commercial system under collaboration with industries is currently underway., kaken
- Grant-in-Aid for Scientific Research (B), 1999, 2001, 11694159, Integrated Machine Learning Workbench for Data Mining, MOTODA Hiroshi; YOSHIDA Tetsuya; WASHIO Takashi, Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, Osaka University, 8700000, 8700000, A new generation of computational techniques and tools is required to support the extraction of useful knowledge from the rapidly growing volumes of data. In this research project we aimed to develop effective methods for feature selection, instance selection and feature construction and integrate them to form a basis of workbench for machine learning and data mining. For feature selection, various performance measures such as distance measure, uncertainty measure, dependency measure, consistency measure and error rate, and various search methods such as heuristic search, complete search and random search were investigated and a design strategy was proposed as to which method to use for which kind of dataset. Further, a new method ABB was proposed that uses consistency measure and performs a very efficient complete search. For instance selection, a new method S^3 Bagging which combines random subsampling and committee learning method was proposed and it was expected that this reduces the amount of data by 90%. For feature construction, two new methods were proposed. One is multi-strategy learning in which graph-base induction GBI that is based on repeated chunking of paired nodes was used as a feature constructor for use in decision tree classifier. Another is to construct new features from association rules. Both were tested against various datasets and conformed effective. All of these are components of the workbench, and we expect that this contributes to mining better knowledge more efficiently., kaken
- 奨励研究(A), 1999, 2000, 11750351, 複数のシステムの間における競合解消戦略の協調的獲得機能の研究, 吉田 哲也, 日本学術振興会, 科学研究費助成事業, 大阪大学, 2100000, 2100000, 本年度は,昨年度に提案したマルチエージェントシステムにおける強化学習手法を人工衛星の自動設計に関する評価実験を通じて検証した.機械学習の分野で研究が進められてきた強化学習の手法において,本研究ではエージェントの自己主張が部分問題の最適化を達成するととらえ,個々の解の評価に応じた報酬(これを解報酬と呼ぶ)を与え,また,エージェント相互の意見のすりあわせが解における部分問題間での不都合を解消しその間での適切なバランスを達成するととらえ,エージェント相互の譲歩に対する報酬(これを調和報酬と呼ぶ)を個々の解の間の差異の程度に応じて与えることで,エージェントに報酬に基づく強化学習を行わている.小型人工衛星の自動設計を構造設計,熱設計,通信設計のサブシステムに分解し,それぞれの部分設計をエージェントに行わせることにより衛星全体の設計を進めていくマルチエージェントシステムをDOS/Vマシン上にSmalltalk言語を用いてインプリメントし,人工衛星のミッション,重量,軌道等を変更して様々な種類の実験を行った.実験から,本研究で提案した2種類の報酬の相対的な大きさの関係をマルチエージェントシステムにおける新たなコントロールパラメーターとして利用し,個々の部分問題を最適化した解の生成を重視する場合には解報酬を相対的に大きく設定し,逆に部分問題間でのバランスのとれた解の生成を重視する場合には調和報酬を相対的に大きく設定することで,システム全体としての目的に応じた適切な競合解消戦略を報酬に基づく強化学習により協調的に獲得させることが可能になることの示唆を得られた., kaken
- 萌芽的研究, 1998, 1999, 10875080, ヒューマンインタラクションにおける創発のモデル化, 西田 正吾; 吉田 哲也; 才脇 直樹, 日本学術振興会, 科学研究費助成事業, 大阪大学, 2000000, 2000000, 人間の相互のインタラクションに注目してみると,人間は他人との対話や協同作業を通じて,本人が気付いていなかった全く新しいアイデアを見つけたり,そのきっかけを得たりすることがしばしば観測される.本年度は,「視点や価値観の異なる他者からの示唆,指摘による発想転換」「議論による理解の深化や新たな問題の発見」に見られるような現象を分析することを通じて,ヒューマンインタラクションにおける創発のモデルを検討した.その第1歩として,異分野の人との対話を通じた知的刺激による創発を実現する方式を考案した.ここでは非線形現象の引き込みや相転移などと対応がつくような創発のモデル化を試みた.ついで,対話を通じた知的刺激を基に,組織におけるヒューマンインタラクションのモデルを検討し,また知的刺激を利用した創発的なメディア生成を行う手法を検討した.組織内の人々の間におけるヒューマンインタラクションを通じて組織全体として創発的な現象が生じることを解明するために,組織におけるヒューマンインタラクションのモデル化に関しての研究を行った.重要な情報とそうでない情報を区別することを通じて組織における適切な意思決定を支援するために,定性的シミュレーションを用いた緊急時における情報フィルタリングの手法を提案した.さらに,組織において上層部の戦略的な意思決定における思惑と現場で直面すべき現実的な戦術との乖離を防ぐために,意図と状況の乖離を表象するインタフェースを開発した., kaken
- Grant-in-Aid for Scientific Research (B), 1997, 1999, 09450159, Research on Cooperative Knowledge Acquisition from Multiple People, NISHIDA Shogo; YOSHIDA Tetsuya; SAIWAKI Naoki, Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, Osaka University, 6400000, 6400000, It is required to integrate the knowledge and skill of multiple people into the unified knowledge. For instance, the motor diagnosis knowledge is usually composed of the knowledge from designers and that from operators. Furthermore, people with electric background and those with mechanical background participate in the motor diagnosis. Thus, it is necessary to provide some methodology which enables the integration of such a scraps of information into the unified knowledge. This research tried to provide a methodology for the cooperative knowledge acquisition from multiple people. First, a method is proposed to detect conceptual difference between people. Based on the method, a communication support method in organization is proposed. Furthermore, a cooperative work support method via the smooth communication is also proposed., kaken
- 特定領域研究(A), 1998, 1998, 10143211, 複数の人間が作成した大規模データベースにおける概念相違発見手法の研究, 西田 正吾; 吉田 哲也; 才脇 直樹, 日本学術振興会, 科学研究費助成事業, 大阪大学, 2500000, 2500000, 本年度は大規模データベースからの知識発見という観点に立ち、高速に(データ数nに対してO(n^3)程度の規模で)複数の人間の間に存在する概念の相違を検出するシステム構築のために、従来から我々が提案してきた手法の計算量解析を行った.一般的に人々は異なった視点や観点を持つため,たとえ同じ事象に対しても異なる概念を持つ可能性がある.概念における相違が存在する場合には同じ事象に対してでも意見を交換して相互理解を深めることが不可能となるため,概念の相違を除くことは背景知識が異なる人々がグループとして共同作業を行う場合に特に重要となる.ユーザーの知識を具体的な決定木として構造化し,概念の差異を決定木間の構造における差異として検出を行うという我々の枠組みにおいて,決定木の生成手法として情報量基準に基づくID3を利用する手法と,「多様な構造化」というコンセプトに基づく遺伝アルゴリズムを利用する手法を提案してきた.遺伝アルゴリズムを利用する手法により概念相違検出精度の向上は達成したものの,非常に長い計算時間を必要とするという新たな問題も明らかになった,このため,大規模データベースからの知識発見という観点に立ち、高速に(データ数nに対してO(n^3)程度の規模で)複数の人間の間に存在する概念の相違を検出するシステム構築のために、遺伝プログラミングを利用した手法の計算量解析を行い,概念相違検出の高速化のための基礎検討を行った., kaken
- 重点領域研究, 1997, 1997, 09230211, 時空間データの知的検索のためのデータ構造の研究, 西田 正吾; 吉田 哲也; 才脇 直樹, 日本学術振興会, 科学研究費助成事業, 大阪大学, 2000000, 2000000, 近年,CG(Computer Graphics)技術やVR(Virtual Reality)技術は著しく進歩しており,コンピュータ内に3次元仮想世界を構築して,自由にウオ-クスルーを行ったり,仮想世界をリアルタイムで変化させてアニメーションを作成したりすることが可能となってきている.本研究では,このような大規模な時空間データを対象にその高速検索のためのデータ構造について検討を行った.具体的には,「ある時空間範囲に含まれるものを探す」「ある点に最も近いものを探す」などの時空間座標に関連した検索を高速に行うことを研究の目的としており,このような機能を実現することにより,3次元仮想世界やアニメーションでの直接的な検索条件の指示およびリアルタイムのレスポンスを実現することを目指した. 3次元空間データや時空間データに関しては,多くの研究が行われており,特に最近「時間」の重要性が認識されてきている.空間データに対しては,空間を再帰的に分割して階層的に管理することによって,データの位置情報に基づいた検索を効率化している一方,医療データ管理システム等に代表される時間情報の管理で扱っているデータは主として文字・数値属性情報であり,図形等のデータに対する空間検索の効率化はあまり考慮されていない. 我々は,従来より木構造に基づいた高速検索のためのデータ構造の研究を進めてきており,地下配管の設備管理への適用検討等も行ってきた.筆者らが提案しているデータ構造(以下,AT構造と呼ぶ)は,時空間データに対して,時間木と空間木を用意しておき,検索範囲に応じて適応的に切り替えることにより,高速な検索を実現しようという方式であり,シミュレーションによりその有効性が確認された.また,複数の状態に対応するAT構造を具体的に検討すると共にシミュレーション実験により性能評価を行った.さらに時間に対して動的に内容が変化するデータに対応するための手法を検討し拡張を試みた., kaken
- 基盤研究(C), 01 Apr. 2021, 31 Mar. 2024, 21K12542, 計算折り紙に基づくスモッキングのパターン生成, 吉田 哲也, 日本学術振興会, 科学研究費助成事業, 奈良女子大学, 4160000, 3200000, 960000, 廉価な計算機やネットワークの普及により，手芸に対してもデジタル・ファブリケーションを活用することが求められている．そこで，本研究では計算折り紙に基づくスモッキングのパターン生成の研究開発を行う．これを実現するために，(1) 計算折り紙に基づくスモッキングのパターンの定式化，(2) 展開図に基づくパターン生成アルゴリズムの開発，(3) 被服の実制作を通じた評価・検証，に取り組む． 上記の目的を実現するために，本年度は，計算折り紙に基づくスモッキングのパターン生成の定式化に取り組んだ．具体的には，下記の項目を実施した． （１） 布の表面の模様の造形に必要なプリーツの折りたたみ操作を折り紙のねじり折りに対する展開図として表現する．このために，折り方の山谷割り当てを指定した折り線が模様の頂点に接続する平面グラフを定義し，折りたたみによる造形を折り線による鏡映変換の合成写像として表現し，展開図に基づくパターンを定義する．さらに，模様ごとに折り線への山谷割り当てが指定された展開図をプリーツを共有するように組み合わせて装飾全体の展開図を生成する配置問題を定式化した． （２） 上記（１）で定式化する，折り紙の展開図に基づくスモッキングのパターンに対して，造形する模様をパターンに写像するための相似比と回転角を導出し，展開図に対する幾何学的変換を用いてパターンを生成するアルゴリズムを開発した． （３） 上記で開発したアルゴリズムを，幾何オブジェクトの操作が可能なGeoGebra を用いてプロトタイプシステムとして実装した．, kaken
- Grant-in-Aid for Scientific Research (B), 01 Apr. 2012, 31 Mar. 2016, 24300049, Transfer Learning based on the structure of Feature Space, YOSHIDA Tetsuya; IMAI Hideyuki, Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, 12350000, 9500000, 2850000, In order to cope with increasing quantity and variety of data, it is important to develop information technology which enables effective reuse of learned knowledge. We have developed a transfer learning method based on the structure of feature space. In the developed method, features are learned from the given data in the source domain, and a graph is constructed based on the learned features. Then, the method tries to preserve the structure of feature space as much as possible between the source domain and the target domain in transfer learning. Under the framework of optimization learning, we have developed a transfer learning algorithm based on multiplicative update rules. The algorithm has been implemented as a prototype system, and experiments over the prototype system were conducted over several benchmark datasets. The results of the experiments indicate the effectiveness of the developed transfer learning method., kaken
- Grant-in-Aid for Scientific Research (A), 2009, 2011, 21240003, Establishing the federation model of smart objects and the creation of its novel applications, TANAKA Yuzuru; YOSHIDA Tetsuya; LUNZER Aran; MIYAZAKI Yutaka; SJOBERGH Jonas; ITO Kimihito, Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, Hokkaido University, 46800000, 36000000, 10800000, This research has focused on three different levels for modeling the proximity-based federation of smart objects. For each of these three levels, i.e., the modeling of the federation mechanism of each smart object, the modeling of the federation mechanism among more than one smart objects, and the modeling of application scenarios, it has proposed three formal models and clarified the mapping among them. They are respectively the port matching model, the graph rewriting model, and the catalytic reaction network model., kaken
- Grant-in-Aid for Scientific Research (C), 2008, 2010, 20500123, A Clustering Method based on graph Coloring for integrating Local and Global Constraints, YOSHIDA Tetsuya, Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, Hokkaido University, 4420000, 3400000, 1020000, We have developed a clustering method based on b-coloring of a graph in Graph Theory. This coloring requires that two kinds of constraints should be satisfied in the coloring of a graph. By regarding one kind of constraint as a local constraint and the other as a global constraint, we have developed a clustering method which reflects both constraints in a unified manner. For a given dataset, we define a graph structure based on the pairwise dissimilarities so that the coloring of the dataset can be conducted over the graph. The developed algorithm conducts the coloring of the graph for conducting the clustering of the dataset. The algorithm has been implemented as a prototype system, and experiments using the prototype system were conducted over several datasets. The results of the experiments indicate the effectiveness of the developed clustering algorithm., kaken