Masashi SUGIYAMA


Stochastic Divergence Minimization for Biterm Topic Models
Zhenghang CUI Issei SATO Masashi SUGIYAMA 
Publication:   
Publication Date: 2018/03/01
Vol. E101-D  No. 3  pp. 668-677
Type of Manuscript:  PAPER
Category: Data Engineering, Web Information Systems
Keyword: 
short texttopic modelbitermstochastic inference algorithm
 Summary | Full Text:PDF(605.7KB)

Computationally Efficient Class-Prior Estimation under Class Balance Change Using Energy Distance
Hideko KAWAKUBO Marthinus Christoffel DU PLESSIS Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2016/01/01
Vol. E99-D  No. 1  pp. 176-186
Type of Manuscript:  PAPER
Category: Artificial Intelligence, Data Mining
Keyword: 
class balance changeclass-prior estimationenergy distance
 Summary | Full Text:PDF(862.6KB)

Direct Density Ratio Estimation with Convolutional Neural Networks with Application in Outlier Detection
Hyunha NAM Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2015/05/01
Vol. E98-D  No. 5  pp. 1073-1079
Type of Manuscript:  PAPER
Category: Artificial Intelligence, Data Mining
Keyword: 
density ratio estimationconvolutional neural networkoutlier detection
 Summary | Full Text:PDF(588.5KB)

Unsupervised Dimension Reduction via Least-Squares Quadratic Mutual Information
Janya SAINUI Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2014/10/01
Vol. E97-D  No. 10  pp. 2806-2809
Type of Manuscript:  LETTER
Category: Artificial Intelligence, Data Mining
Keyword: 
unsupervised dimension reductionquadratic mutual informationleast-squares density differenceEpanechnikov kernelhyperparameter tuning
 Summary | Full Text:PDF(503.6KB)

Constrained Least-Squares Density-Difference Estimation
Tuan Duong NGUYEN Marthinus Christoffel DU PLESSIS Takafumi KANAMORI Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2014/07/01
Vol. E97-D  No. 7  pp. 1822-1829
Type of Manuscript:  PAPER
Category: Artificial Intelligence, Data Mining
Keyword: 
density differenceasymptotic varianceL2-distancebiasclass-balance changetwo-sample homogeneity test
 Summary | Full Text:PDF(554.4KB)

Tree-Based Ensemble Multi-Task Learning Method for Classification and Regression
Jaak SIMM Ildefons MAGRANS DE ABRIL Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2014/06/01
Vol. E97-D  No. 6  pp. 1677-1681
Type of Manuscript:  LETTER
Category: Pattern Recognition
Keyword: 
machine learningmulti-task learningtree-based methodsensemble methods
 Summary | Full Text:PDF(180.3KB)

Class Prior Estimation from Positive and Unlabeled Data
Marthinus Christoffel DU PLESSIS Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2014/05/01
Vol. E97-D  No. 5  pp. 1358-1362
Type of Manuscript:  LETTER
Category: Artificial Intelligence, Data Mining
Keyword: 
class-prior changeoutlier detectionpositive and unlabeled learningdivergence estimationpearson divergence
 Summary | Full Text:PDF(449.9KB)

Computationally Efficient Estimation of Squared-Loss Mutual Information with Multiplicative Kernel Models
Tomoya SAKAI Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2014/04/01
Vol. E97-D  No. 4  pp. 968-971
Type of Manuscript:  LETTER
Category: Fundamentals of Information Systems
Keyword: 
squared-loss mutual informationleast-squares mutual informationdensity ratio estimationmultiplicative kernel modelsindependence test
 Summary | Full Text:PDF(563.4KB)

Direct Approximation of Quadratic Mutual Information and Its Application to Dependence-Maximization Clustering
Janya SAINUI Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2013/10/01
Vol. E96-D  No. 10  pp. 2282-2285
Type of Manuscript:  LETTER
Category: Artificial Intelligence, Data Mining
Keyword: 
quadratic mutual informationleast-square density-difference estimationinformation theoretic clustering
 Summary | Full Text:PDF(219.5KB)

Computationally Efficient Multi-Label Classification by Least-Squares Probabilistic Classifiers
Hyunha NAM Hirotaka HACHIYA Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2013/08/01
Vol. E96-D  No. 8  pp. 1871-1874
Type of Manuscript:  LETTER
Category: Fundamentals of Information Systems
Keyword: 
multi-label classificationleast-squares probabilistic classifier
 Summary | Full Text:PDF(185.7KB)

Feature Selection via 1-Penalized Squared-Loss Mutual Information
Wittawat JITKRITTUM Hirotaka HACHIYA Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2013/07/01
Vol. E96-D  No. 7  pp. 1513-1524
Type of Manuscript:  PAPER
Category: Pattern Recognition
Keyword: 
feature selection1-regularizationsquared-loss mutual informationdensity-ratio estimationdimensionality reduction
 Summary | Full Text:PDF(979.6KB)

Artist Agent: A Reinforcement Learning Approach to Automatic Stroke Generation in Oriental Ink Painting
Ning XIE Hirotaka HACHIYA Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2013/05/01
Vol. E96-D  No. 5  pp. 1134-1144
Type of Manuscript:  PAPER
Category: Artificial Intelligence, Data Mining
Keyword: 
painterly renderingstroke-based renderingreinforcement learningpolicy gradient
 Summary | Full Text:PDF(5.1MB)

Winning the Kaggle Algorithmic Trading Challenge with the Composition of Many Models and Feature Engineering
Ildefons MAGRANS DE ABRIL Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2013/03/01
Vol. E96-D  No. 3  pp. 742-745
Type of Manuscript:  LETTER
Category: Artificial Intelligence, Data Mining
Keyword: 
Kaggle challengemodel architectureboostingfeature selectionhigh frequency tradingliquidity shockmaximal information coefficient
 Summary | Full Text:PDF(151.5KB)

Multi-Task Approach to Reinforcement Learning for Factored-State Markov Decision Problems
Jaak SIMM Masashi SUGIYAMA Hirotaka HACHIYA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2012/10/01
Vol. E95-D  No. 10  pp. 2426-2437
Type of Manuscript:  PAPER
Category: Artificial Intelligence, Data Mining
Keyword: 
reinforcement learningmulti-task learningtransfer learningfactored state modelsMarkov decision process
 Summary | Full Text:PDF(508.1KB)

On Kernel Parameter Selection in Hilbert-Schmidt Independence Criterion
Masashi SUGIYAMA Makoto YAMADA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2012/10/01
Vol. E95-D  No. 10  pp. 2564-2567
Type of Manuscript:  LETTER
Category: Artificial Intelligence, Data Mining
Keyword: 
Hilbert-Schmidt independence criterionleast-squares mutual informationcross-validationGaussian kernel
 Summary | Full Text:PDF(127.7KB)

Early Stopping Heuristics in Pool-Based Incremental Active Learning for Least-Squares Probabilistic Classifier
Tsubasa KOBAYASHI Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2012/08/01
Vol. E95-D  No. 8  pp. 2065-2073
Type of Manuscript:  PAPER
Category: Artificial Intelligence, Data Mining
Keyword: 
pool-based incremental active learningearly stoppingleast-squares probabilistic classifiersemi-supervised learning
 Summary | Full Text:PDF(2MB)

FOREWORD
Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2011/10/01
Vol. E94-D  No. 10  pp. 1845-1845
Type of Manuscript:  FOREWORD
Category: 
Keyword: 
 Summary | Full Text:PDF(57.4KB)

Improving the Accuracy of Least-Squares Probabilistic Classifiers
Makoto YAMADA Masashi SUGIYAMA Gordon WICHERN Jaak SIMM 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2011/06/01
Vol. E94-D  No. 6  pp. 1337-1340
Type of Manuscript:  LETTER
Category: Pattern Recognition
Keyword: 
least-squares probabilistic classifierkernel logistic regressiondensity ratioPASCAL VOC 2010freesound
 Summary | Full Text:PDF(114.6KB)

Least-Squares Independence Test
Masashi SUGIYAMA Taiji SUZUKI 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2011/06/01
Vol. E94-D  No. 6  pp. 1333-1336
Type of Manuscript:  LETTER
Category: Artificial Intelligence, Data Mining
Keyword: 
independence testdensity ratio estimationunconstrained least-squares importance fittingsquared-loss mutual information
 Summary | Full Text:PDF(161.5KB)

Lighting Condition Adaptation for Perceived Age Estimation
Kazuya UEKI Masashi SUGIYAMA Yasuyuki IHARA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2011/02/01
Vol. E94-D  No. 2  pp. 392-395
Type of Manuscript:  LETTER
Category: Image Recognition, Computer Vision
Keyword: 
face recognitionage estimationcovariate shift adaptationlighting condition changeKullback-Leibler importance estimation procedureimportance-weighted regularized least-squares
 Summary | Full Text:PDF(159.1KB)

Superfast-Trainable Multi-Class Probabilistic Classifier by Least-Squares Posterior Fitting
Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2010/10/01
Vol. E93-D  No. 10  pp. 2690-2701
Type of Manuscript:  Special Section PAPER (Special Section on Data Mining and Statistical Science)
Category: 
Keyword: 
probabilistic classificationkernel logistic regressionclass-posterior probabilitysquared-loss
 Summary | Full Text:PDF(1.2MB)

A Semi-Supervised Approach to Perceived Age Prediction from Face Images
Kazuya UEKI Masashi SUGIYAMA Yasuyuki IHARA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2010/10/01
Vol. E93-D  No. 10  pp. 2875-2878
Type of Manuscript:  LETTER
Category: Image Recognition, Computer Vision
Keyword: 
perceived age estimationactive learningweighted regressionsemi-supervised learningmanifold regularizationhuman age perception
 Summary | Full Text:PDF(199KB)

Direct Importance Estimation with a Mixture of Probabilistic Principal Component Analyzers
Makoto YAMADA Masashi SUGIYAMA Gordon WICHERN Jaak SIMM 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2010/10/01
Vol. E93-D  No. 10  pp. 2846-2849
Type of Manuscript:  LETTER
Category: Fundamentals of Information Systems
Keyword: 
importanceKullback-Leibler importance estimation proceduremixture of probabilistic principal component analyzersexpectation-maximization algorithm
 Summary | Full Text:PDF(121.1KB)

FOREWORD
Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2010/10/01
Vol. E93-D  No. 10  pp. 2671-2671
Type of Manuscript:  FOREWORD
Category: 
Keyword: 
 Summary | Full Text:PDF(56.1KB)

Least Absolute Policy Iteration--A Robust Approach to Value Function Approximation
Masashi SUGIYAMA Hirotaka HACHIYA Hisashi KASHIMA Tetsuro MORIMURA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2010/09/01
Vol. E93-D  No. 9  pp. 2555-2565
Type of Manuscript:  PAPER
Category: Artificial Intelligence, Data Mining
Keyword: 
reinforcement learningvalue function approximationleast-squares policy iterationoutlier1-loss functionlinear programming
 Summary | Full Text:PDF(2.1MB)

Spectral Methods for Thesaurus Construction
Nobuyuki SHIMIZU Masashi SUGIYAMA Hiroshi NAKAGAWA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2010/06/01
Vol. E93-D  No. 6  pp. 1378-1385
Type of Manuscript:  Special Section PAPER (Special Section on Info-Plosion)
Category: Natural Language Processing
Keyword: 
synonym acquisitionsynonym extractionthesaurusspectral clusteringgraph laplacian
 Summary | Full Text:PDF(257.7KB)

Theoretical Analysis of Density Ratio Estimation
Takafumi KANAMORI Taiji SUZUKI Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences
Publication Date: 2010/04/01
Vol. E93-A  No. 4  pp. 787-798
Type of Manuscript:  PAPER
Category: Algorithms and Data Structures
Keyword: 
density ratio estimationdensity estimationlogistic regressionasymptotic analysisGaussian assumption
 Summary | Full Text:PDF(246.7KB)

Least-Squares Conditional Density Estimation
Masashi SUGIYAMA Ichiro TAKEUCHI Taiji SUZUKI Takafumi KANAMORI Hirotaka HACHIYA Daisuke OKANOHARA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2010/03/01
Vol. E93-D  No. 3  pp. 583-594
Type of Manuscript:  PAPER
Category: Pattern Recognition
Keyword: 
conditional density estimationmultimodalityheteroscedastic noisedirect density ratio estimationtransition estimation
 Summary | Full Text:PDF(493.3KB)

Direct Importance Estimation with Gaussian Mixture Models
Makoto YAMADA Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2009/10/01
Vol. E92-D  No. 10  pp. 2159-2162
Type of Manuscript:  LETTER
Category: Pattern Recognition
Keyword: 
importance weightKLIEPGaussian mixture modelsEM algorithm
 Summary | Full Text:PDF(135.4KB)

Recent Advances and Trends in Large-Scale Kernel Methods
Hisashi KASHIMA Tsuyoshi IDE Tsuyoshi KATO Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2009/07/01
Vol. E92-D  No. 7  pp. 1338-1353
Type of Manuscript:  INVITED PAPER (Special Section on Large Scale Algorithms for Learning and Optimization)
Category: 
Keyword: 
kernel methodssupport vector machineskernel tricklow-rank approximationoptimizationstructured data
 Summary | Full Text:PDF(308KB)

On Computational Issues of Semi-Supervised Local Fisher Discriminant Analysis
Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2009/05/01
Vol. E92-D  No. 5  pp. 1204-1208
Type of Manuscript:  LETTER
Category: Artificial Intelligence and Cognitive Science
Keyword: 
dimensionality reductionsemi-supervised learningsemi-supervised local Fisher discriminant analysissparsitygeneralized eigenvalue problem
 Summary | Full Text:PDF(94.1KB)

Approximating the Best Linear Unbiased Estimator of Non-Gaussian Signals with Gaussian Noise
Masashi SUGIYAMA Motoaki KAWANABE Gilles BLANCHARD Klaus-Robert MULLER 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2008/05/01
Vol. E91-D  No. 5  pp. 1577-1580
Type of Manuscript:  LETTER
Category: Pattern Recognition
Keyword: 
signal denoisingbest linear unbiased estimator (BLUE)non-Gaussian component analysis (NGCA)Gaussian noise
 Summary | Full Text:PDF(159.7KB)

A New Meta-Criterion for Regularized Subspace Information Criterion
Yasushi HIDAKA Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2007/11/01
Vol. E90-D  No. 11  pp. 1779-1786
Type of Manuscript:  PAPER
Category: Pattern Recognition
Keyword: 
supervised learninggeneralization capabilitymodel selectionunbiased estimatorregularized subspace information criterion
 Summary | Full Text:PDF(208.3KB)

Analytic Optimization of Adaptive Ridge Parameters Based on Regularized Subspace Information Criterion
Shun GOKITA Masashi SUGIYAMA Keisuke SAKURAI 
Publication:   IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences
Publication Date: 2007/11/01
Vol. E90-A  No. 11  pp. 2584-2592
Type of Manuscript:  PAPER
Category: Neural Networks and Bioengineering
Keyword: 
supervised learninggeneralization errormodel selectionregularized subspace information criterion
 Summary | Full Text:PDF(208.7KB)

Generalization Error Estimation for Non-linear Learning Methods
Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences
Publication Date: 2007/07/01
Vol. E90-A  No. 7  pp. 1496-1499
Type of Manuscript:  LETTER
Category: Neural Networks and Bioengineering
Keyword: 
supervised learninggeneralization capabilitymodel selectionsubspace information criterionnon-linear learning
 Summary | Full Text:PDF(92.6KB)

Analytic Optimization of Shrinkage Parameters Based on Regularized Subspace Information Criterion
Masashi SUGIYAMA Keisuke SAKURAI 
Publication:   IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences
Publication Date: 2006/08/01
Vol. E89-A  No. 8  pp. 2216-2225
Type of Manuscript:  PAPER
Category: Neural Networks and Bioengineering
Keyword: 
supervised learninggeneralization capabilitymodel selectionshrinkage estimatorregularized subspace information criterion
 Summary | Full Text:PDF(371.9KB)

Constructing Kernel Functions for Binary Regression
Masashi SUGIYAMA Hidemitsu OGAWA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2006/07/01
Vol. E89-D  No. 7  pp. 2243-2249
Type of Manuscript:  PAPER
Category: Pattern Recognition
Keyword: 
supervised learningregressionkernel methodskernel functionsKarhunen-Loeve expansionprincipal component analysisbinary regressionGaussian kernel
 Summary | Full Text:PDF(243.7KB)

Active Learning with Model Selection -- Simultaneous Optimization of Sample Points and Models for Trigonometric Polynomial Models
Masashi SUGIYAMA Hidemitsu OGAWA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2003/12/01
Vol. E86-D  No. 12  pp. 2753-2763
Type of Manuscript:  PAPER
Category: Pattern Recognition
Keyword: 
supervised learninggeneralization capabilityactive learningmodel selectiontrigonometric polynomial space
 Summary | Full Text:PDF(966.8KB)

Improving Precision of the Subspace Information Criterion
Masashi SUGIYAMA 
Publication:   IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences
Publication Date: 2003/07/01
Vol. E86-A  No. 7  pp. 1885-1895
Type of Manuscript:  PAPER
Category: Neural Networks and Bioengineering
Keyword: 
machine learninggeneralization capabilitysubspace information criteriongeneralized inverseunbiased estimatorvariance
 Summary | Full Text:PDF(1.1MB)

Incremental Construction of Projection Generalizing Neural Networks
Masashi SUGIYAMA Hidemitsu OGAWA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2002/09/01
Vol. E85-D  No. 9  pp. 1433-1442
Type of Manuscript:  PAPER
Category: Biocybernetics, Neurocomputing
Keyword: 
generalization capabilityincremental learningpseudo biorthogonal basis (PBOB)projection learningprojection generalizing neural network
 Summary | Full Text:PDF(484.8KB)

Active Learning for Optimal Generalization in Trigonometric Polynomial Models
Masashi SUGIYAMA Hidemitsu OGAWA 
Publication:   IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences
Publication Date: 2001/09/01
Vol. E84-A  No. 9  pp. 2319-2329
Type of Manuscript:  PAPER
Category: Algorithms and Data Structures
Keyword: 
supervised learningactive learninggeneralization capabilitytrigonometric polynomial spacepseudo orthogonal bases
 Summary | Full Text:PDF(399.2KB)

Subspace Information Criterion for Image Restoration--Optimizing Parameters in Linear Filters
Masashi SUGIYAMA Daisuke IMAIZUMI Hidemitsu OGAWA 
Publication:   IEICE TRANSACTIONS on Information and Systems
Publication Date: 2001/09/01
Vol. E84-D  No. 9  pp. 1249-1256
Type of Manuscript:  PAPER
Category: Image Processing, Image Pattern Recognition
Keyword: 
image restorationmean squared errorsubspace information criterionmoving-average filtermodel selection
 Summary | Full Text:PDF(6.6MB)