Graph Classification Based on Optimizing Graph Spectra.

Development of Data Mining Platform MUSASHI Towards Service Computing.

Discovery of Exogenous Variables in Data with More Variables Than Observations.
Abstract. Many statistical methods have been proposed to estimate causal models in classical situations with fewer variables than observations. However, modern datasets including gene expression data increase the needs of high-dimensional causal modeling in challenging situations with orders of magnitude more variables than observations. In this paper, we propose a method to find exogenous variables in a linear non-Gaussian causal model, which requires much smaller sample sizes than conventional methods and works even when orders of magnitude more variables than observations. Exogenous variables work as triggers that activate causal chains in the model, and their identification leads to more efficient experimental designs and better understanding of the causal mechanism. We present experiments with artificial data and real-world gene expression data to evaluate the method.
Stationary Subspace Analysis as a Generalized Eigenvalue Problem.

Use of Prior Knowledge in a Non%2DGaussian Method for Learning Linear Structural Equation Models.

GTRACE2 Improving Performance Using Labeled Union Graphs.

Mining Frequent Graph Sequence Patterns Induced by Vertices.

GroupLiNGAM Linear non%2DGaussian acyclic models for sets of variables

GTRACE Mining Frequent Subsequences from Graph Sequences.

GTRACE Mining Frequent Subsequences from Graph Sequences.

Best papers from the 12th Pacific%2DAsia conference on knowledge discovery and data mining %28PAKDD2008%29.

New Frontiers in Applied Data Mining

Advances in Machine Learning
Game-playing applications offer various challenges for machine learning, including opening book learning, learning of evaluation functions, player modeling, and others. In this paper, we briefly highlight the most important achievements of this field and summarize some important recent advance.  
Optimization of Budget Allocation for TV Advertising.

Special Issue on Data%2DMining and Statistical Science.

Advances in Knowledge Discovery and Data Mining

A Fast Method to Mine Frequent Subsequences from Graph Sequence Data.

A Bank Run Model in Financial Crises.

A Range Query Approach for High Dimensional Euclidean Space Based on EDM Estimation.

Pruning Strategies Based on the Upper Bound of Information Gain for Discriminative Subgraph Mining.

A Range Query Approach for High Dimensional Euclidean Space Based on EDM Estimation.

DryadeParent%2C An Efficient and Robust Closed Attribute Tree Mining Algorithm.

Modeling dynamic substate chains among massive states.

New Frontiers in Artificial Intelligence
Abstract. In this paper, we discuss a viewpoint to regard individuals in a society as cognitive agents having internal dynamics, in order to study the dynamic nature of social structures. Internal dynamics is the autonomous changes of an agent’s internal states that govern his/her behavior. We first discuss the benefit of introducing internal dynamics into a model of humans and the dynamics of society. Then we propose a simple recurrent network with self-influential connection (SRN-SIC) as a model of an agent with internal dynamics. We report the results of our simulation in which the agents play a minority game. In the simulation, we observe the dynamics of the game as a macro structure itinerating among various dynamical states such as fixed points and periodic motions via aperiodic motions. This itinerant change of the macro structures is shown to be induced by the internal dynamics of the agents.
Emerging Technologies in Knowledge Discovery and Data Mining

Communicability Criteria of Law Equations Discovery.

Robust Unsupervised and Semisupervised Bounded C%2DSupport Vector Machines.

International Workshop on Risk Informatics %28RI2007%29.

DIGDAG%2C a First Algorithm to Mine Closed Frequent Embedded Sub%2DDAGs.


Applications eligible for data mining.


New Frontiers in Artificial Intelligence
Abstract. In this paper, we discuss a viewpoint to regard individuals in a society as cognitive agents having internal dynamics, in order to study the dynamic nature of social structures. Internal dynamics is the autonomous changes of an agent’s internal states that govern his/her behavior. We first discuss the benefit of introducing internal dynamics into a model of humans and the dynamics of society. Then we propose a simple recurrent network with self-influential connection (SRN-SIC) as a model of an agent with internal dynamics. We report the results of our simulation in which the agents play a minority game. In the simulation, we observe the dynamics of the game as a macro structure itinerating among various dynamical states such as fixed points and periodic motions via aperiodic motions. This itinerant change of the macro structures is shown to be induced by the internal dynamics of the agents.
Modeling Dynamic Substate Chains among Massive States for Prediction.

A Method to Search ARX Model Orders and Its Application to Sales Dynamics Analysis.

Risk Mining %2D Overview.

Analysis on a Relation Between Enterprise Profit and Financial State by Using Data Mining Techniques.

Constructing Decision Trees for Graph%2DStructured Data by Chunkingless Graph%2DBased Induction.

Extracting Discriminative Patterns from Graph Structured Data Using Constrained Search.

A study on rough set%2Daided feature selection for automatic web%2Dpage classification.

SCALETRACK A System to Discover Dynamic Law Equations Containing Hidden States and Chaos.

Efficient Mining of High Branching Factor Attribute Trees.

Discovering Time Differential Law Equations Containing Hidden State Variables and Chaotic Dynamics.

Mining Quantitative Frequent Itemsets Using Adaptive Density%2DBased Subspace Clustering.

Discovering Time Differential Law Equations Containing Hidden State Variables and Chaotic Dynamics.

Mutagenicity Risk Analysis by Using Class Association Rules.

Cl%2DGBI A Novel Approach for Extracting Typical Patterns from Graph%2DStructured Data.

Deriving Class Association Rules Based on Levelwise Subspace Clustering.

Memory Management of Density%2DBased Spam Detector.

A Framework of Numerical Basket Analysis.

A General Framework for Mining Frequent Subgraphs from Labeled Graphs.

Advances in Mining Graphs%2C Trees and Sequences.

Enhancing the plausibility of law equation discovery through cross check among multiple scale%2Dtype%2Dbased models.

Multi%2Dstructure Information Retrieval Method Based on Transformation Invariance.

Density%2Dbased spam detector.

Consumer Behavior Analysis by Graph Mining Technique.
Abstract. In this paper we discuss how graph mining system is applied to sales transaction data so as to understand consumer behavior. First, existing research of consumer behavior analysis for sequential purchase pattern is reviewed. Then we propose to represent the complicated customer purchase behavior by a directed graph retaining temporal information in a purchase sequence and apply a graph mining technique to analyze the frequent occurring patterns. In this paper we demonstrate through the case of healthy cooking oil analysis how graph mining technology helps us understand complex purchase behavior. 1
Using a Hash%2DBased Method for Apriori%2DBased Graph Mining.



Extracting Diagnostic Knowledge from Hepatitis Dataset by Decision Tree Graph%2DBased Induction.

Data Mining Oriented CRM Systems Based on MUSASHI C%2DMUSASHI.

Performance Evaluation of Decision Tree Graph%2DBased Induction.

Development of Generic Search Method Based on Transformation Invariance.

Classifier Construction by Graph%2DBased Induction for Graph%2DStructured Data.

Complete Mining of Frequent Patterns from Graphs Mining Graph Data.
Abstract. Basket Analysis, which is a standard method for data mining, derives frequent itemsets from database. However, its mining ability is limited to transaction data consisting of items. In reality, there are many applications where data are described in a more structural way, e.g. chemical compounds and Web browsing history. There are a few approaches that can discover characteristic patterns from graph-structured data in the field of machine learning. However, almost all of them are not suitable for such applications that require a complete search for all frequent subgraph patterns in the data. In this paper, we propose a novel principle and its algorithm that derive the characteristic patterns which frequently appear in graphstructured data. Our algorithm can derive all frequent induced subgraphs from both directed and undirected graph structured data having loops (including self-loops) with labeled or unlabeled nodes and links. Its performance is evaluated through the applications to Web browsing pattern analysis and chemical carcinogenesis analysis.
State of the art of graph%2Dbased data mining.

Mining Patterns from Structured Data by Beam%2DWise Graph%2DBased Induction.
Graph-Based Induction (GBI) extracts typical patterns from  graph data by stepwise pair expansion (pairwise chunking). It is very e#-  cient because of its greedy search strategy but at the same time it su#ers  from the incompleteness of search. Improvement is made on its search  capability without imposing much computational complexity by 1) incorporating  a beam search, 2) using a di#erent evaluation function to extract  patterns that are more discriminatory than those simply occurring  frequently, and 3) adopting canonical labeling to enumerate identical patterns  accurately. This new algorithm, now called Beam-wise GBI, B-GBI  for short, was tested against the promoter dataset from UCI repository  and shown successful in extracting discriminatory substructures. E#ect  of beam width on the number of discovered attributes and predictive  accuracy was evaluated. The best result obtained by this approach was  better than the previously best known result. B-GBI was then applied to  a real-world data, Hepatitis dataset provided by Chiba University. Our  very preliminary results indicate that B-GBI can actually handle graphs  with a few thousands nodes and extract discriminatory patterns.
Adaptive Ripple Down Rules Method based on Minimum Description Length Principle.

Extension of the RDR Method That Can Adapt to Environmental Changes and Acquire Knowledge from Both Experts and Data.

Case Generation Method for Constructing an RDR Knowledge Base.

Knowledge Discovery from Structured Data by Beam%2DWise Graph%2DBased Induction.

Inductive Thermodynamics from Time Series Data Analysis.

Toward the Discovery of First Principle Based Scientific Law Equations.

Graph%2Dbased induction and its applications.

Attribute Generation Based on Association Rules.

New Frontiers in Artificial Intelligence
Abstract. In this paper, we discuss a viewpoint to regard individuals in a society as cognitive agents having internal dynamics, in order to study the dynamic nature of social structures. Internal dynamics is the autonomous changes of an agent’s internal states that govern his/her behavior. We first discuss the benefit of introducing internal dynamics into a model of humans and the dynamics of society. Then we propose a simple recurrent network with self-influential connection (SRN-SIC) as a model of an agent with internal dynamics. We report the results of our simulation in which the agents play a minority game. In the simulation, we observe the dynamics of the game as a macro structure itinerating among various dynamical states such as fixed points and periodic motions via aperiodic motions. This itinerant change of the macro structures is shown to be induced by the internal dynamics of the agents.
Discovering Admissible Simultaneous Equation Models from Observed Data.
Abstract. Conventional work on scienti c discovery such as BACON derives empirical law equations from experimental data. In recent years, SDS introducing mathematical admissibility constraints has been proposed to discover rst principle based law equations, and it has been further extended to discover law equations from passively observed data. Furthermore, SSF has been proposed to discover the structure of a simultaneous equation model representing an objective process through experiments. In this paper, SSF is extended to discover the structure of a simultaneous equation model from passively observed data, and is combined with the extended SDS to discover a quantitative simultaneous equation model re ecting the rst principle. 1
S3Bagging Fast Classifier Induction Method with Subsampling and Bagging.

JSAI KDD Challenge 2001 JKDD01.

Knowledge Acquisition from Both Human Expert and Data.

Basket Analysis on Meningitis Data.

Knowledge Acquisition from Both Human Expert and Data.

Automatic Web%2DPage Classification by Using Machine Learning Methods.
Abstract. This paper describes automatic Web-page classification by using machine learning methods. Recently, the importance of portal site services is increasing including the search engine function on World Wide Web. Especially, the portal site such as for Yahoo! service which hierarchically classifies Web-pages into many categories is becoming popular. However, the classification of Web-page into each category exclusively relies on man power which costs much time and care. To alleviate this problem, we propose techniques to generate attributes by using cooccurrence analysis and to classify Web-page automatically based on machine learning. We apply these techniques to Web-pages on Yahoo! JAPAN and construct decision trees which determine appropriate category for each Web-page. The performance of this proposed method is evaluated in terms of error rate, recall, and precision. The experimental evaluation demonstrates that this method provides high accuracy with the classification of Web-page into top level categories on Yahoo! JAPAN. 1
A Description Length%2DBased Decision Criterion for Default Knowledge in the Ripple Down Rules Method.

Nonequilibrium Thermodynamics from Time Series Data Analysis.

Graph%2DBased Induction for General Graph Structured Data and Its Application to Chemical Compound Data.

Enhancing the Plausibility of Law Equation Discovery.
After the pioneering work of the BACON system,  the study in the field of scientific discovery  has been directed to the discovery of  more plausible law equations to represent the  first principles underlying objective systems.  The state of the art has only succeeded in  a weak sense that the soundness, the reproducibility  and the mathematical admissibility  of the candidates hold within the experimental  measurements. The plausibility should  be checked for various objects and/or measurements  sharing the common first principles,  and only the equations having sufficient  generality should be retained. In this paper,  anovel principle and an algorithm are proposed  to predict some mathematically admissible  and consistent equation formulae for a  newly given set of quantities from the candidate  law equations obtained for another set of  quantities in advance. The soundness and the  reproducibility of the predicted equations are  confirmed through the measurements. The  law equatio...
Extension of Graph%2DBased Induction for General Graph Structured Data.

An Apriori%2DBased Algorithm for Mining Frequent Substructures from Graph Data.
. This paper proposes a novel approach named AGM to efficiently  mine the association rules among the frequently appearing substructures  in a given graph data set. A graph transaction is represented  by an adjacency matrix, and the frequent patterns appearing in the matrices  are mined through the extended algorithm of the basket analysis.  Its performance has been evaluated for the artificial simulation data and  the carcinogenesis data of Oxford University and NTP. Its high efficiency  has been confirmed for the size of a real-world problem. . . .  1 Introduction  Mining knowledge from structured data is a major research topic in recent data mining study. &#034;Graph structure&#034; is one of the representatives of the structured data since it frequently appears in real-world data such as web links and chemical compounds structures. In the field of chemistry, CASE and MultiCASE systems have been often used to discover characteristic substructures of chemical compounds [8], [9]. Though these s...
%60%60Thermodynamics%27%27 from Time Series Data Analysis.

Derivation of the Topology Structure from Massive Graph Data.

Graph%2DBased Induction for General Graph Structured Data.

Discovering Admissible Model Equations from Observed Data Based on Scale%2DTypes and Identity Constrains.

A Data Pre%2Dprocessing Method Using Association Rules of Attributes for Improving Decision Tree.

Characterization of Default Knowledge in Ripple Down Rules Method.
&#034;Ripple Down Rules (RDR)&#034; Method is one of the promising approaches to directly acquire and encode knowledge from human experts. It requires data to be supplied incrementally to the knowledge-base being constructed and new piece of knowledge is added as an exception to the existing knowledge. Because of this patching principle, the knowledge acquired strongly depends on what is given as the default knowledge. Further, data are often noisy and we want the RDR noise resistant. This paper reports experimental results about the effect of the selection of default knowledge and the amount of noise in data on the performance of RDR using a simulated expert. The best default knowledge is characterized as the class knowledge that maximizes the minimum description length to encode rules and misclassified cases. The effect of noise is sensitive at an earlier stage of knowledge acquisition where its performance strongly depends on the selection of default knowledge, but RDR eventually converges to its stable performance although the size of the knowledge base is strongly affected by the chosen default knowledge.
Basket Analysis for Graph Structured Data.

Discovering Admissible Simultaneous Equations of Large Scale Systems.

Development of SDS2 Smart Discovery System for Simultaneous Equation Systems.

Mining Association Rules for Estimation and Prediction.

Discovery of first%2Dprinciple equations based on scale%2Dtype%2Dbased and data%2Ddriven reasoning.

Discovering Admissible Models of Complex Systems Based on Scale%2DTypes and Idemtity Constraints.

A New Approach to Quantitative and Credible Diagnosis for Multiple Faults of Components and Sensors.

A Study of Organizational Learning in Multi%2DAgent Sytems.

A History%2DOriented Envisioning Method.

Real Applications on the New Parallel System NEC Cenju%2D3.

