Faculty and Research

Faculty and Research

  • Please see
Campus Plan Web Service

Department of Advanced Data Science

Specialty Courses Keywords
Shiro Ikeda
Both applied and theoretical studies are important in statistics. For the applied research, my research deals with the analysis of real data, especially astronomical data. It covers a wide range of problems, from the imaging in radio interferometry, to the development of signal processing methods for observations at various wavelengths, from X-rays to optical observations, to statistical analysis of cosmology. For theoretical research, we deal with fundamental problems in signal processing and information theory based on information geometry. Special Topics in Signal Processing Astronomical data analysis, Information geometry, Signal processing
Hideitsu Hino
The goal of this course is to introduce mathematical engineering, especially machine learning and data analysis. Students will be instructed to have an intuitive and theoretical explanation of why a certain machine learning algorithm works. Through the research guidance, I try to make the students being able to formulate and solve many realistic problems mathematically. Special Topics in Statiatical Modeling Mathematical engineering, Machine learning, Information geometry
Kenji Fukumizu
This course discusses theoretical and practical aspects of statistical maechine learning and mathematical statistics, aiming at directing research projects of students. Examples of the topics include (1) Kernel methods with positive definite kernels and reproducing kernel Hilbert spaces, (2) Mathematical theory and methods of deep learning, (3) Data analysis methods with geometric features such as topological data analysis. The course focuses new methodologies from broad mathematical viewpoints such as function analysis, geometry, and algebra. Statistical Machine Learning Machine Learning, Deep Learning, Kernel Methods, Topological Data Analysis
Ryo Yoshida
This course covers a wide range of statistical methods in real data analysis. Starting from an education program for machine learning, deep learning, Bayesian inference, experimental design, and R/Python programming, students will systematically learn the methodology of data science through various applications to real problems. For students who are interested in materials informatics and bioinformatics, research guidance will be provided while following up cutting-edge research and technologies in these fields. Application in Data Science Machine learning, materials informatics, bioinformatics
Stephen Wu
From understanding the basic concept to hands-on programming experience, I will provide research guidance to develop and implement statistical methods motivated from real world problems, such as polymer design, molecular dynamics, structural health monitoring, soil property prediction, earthquake early warning, etc. Major statistical methods: Bayesian modeling, uncertainty quantification, sequential Monte Carlo methods, transfer learning Major programming language: Python, Matlab, R Bayesian uncertainty quantification for engineering applications Bayesian Inference, Machine Learning, Mathematical Engineering
Tasuku Soma
Theory and applications of combinatorial ptimization. Specific research topics include (1) Algorithm design for combinatorial optimization problems with submodularity, (2) Combinatorial optimization algorithms with matrix methods and combinatorial approaches to algebraic problems, and (3) Applications of optimization to statistics and machine learning, e.g., optimization with uncertainty and online optimization. Submodular optimization and its applications in machine learning, Linear algebra in combinatorial optimization and algorithm design, Optimization under uncertainty
Ching-Pei Lee
Research study on either nonlinear optimization or large-scale machine learning. For optimization, the topics include theory, algorithms, and implementation, including, but not limited to, those related to nonsmooth regularized optimization and their applications in various fields. For machine learning, we mainly focus on making the training more efficient in various tasks, including federated learning, deep learning, as well as other potential modern machine learning models. Nonlinear optimization, continuous optimization, large-scale machine learning

(Assistant Professor) Yoshihiro Hayashi / Le Thanh Tam

Department of Fundamental Statistical Mathematics

Specialty Courses Keywords
Satoshi Ito
We will study optimization theory and its applications. Specific topics include (1) theory and algorithms of continuous optimization including infinite-dimensional optimization and hierarchical optimization as well as related functional/numerical analysis; (2) systems design under uncertainty with robust optimization, semi-infinite programming, two-level programming and stochastic programming; (3) real-world applications of optimization in the fields of control engineering, signal processing, finance, sports and others. Systems Optimization Infinite-dimensional Optimization, Hierarchical Optimization, Uncertainty
Yukito Iba
I am mostly interested in computational methods for highly multivariate, non-Gaussian (or discrete) probability distributions, as well as their applications to statistics, statistical physics, engineering, machine learning, and combinatorics. Specifically, various applications of Markov chain Monte Carlo and sequential Monte Carlo, are fields of current interest. I am also interested in model selection using MCMC outputs. Special Topics in Modeling 2 Markov chain Monte Carlo, Sequential Monte Carlo, Model selection
Takeshi Emura
Mainly teach multivariate survival analysis methods, but also any statistical methods in general. For example, we try to establish statistical methods for dependence modeling between survival endpoints based on copulas. Students will acquire techniques to understand the structure of complex sampling schemes such as "censoring", "truncation", and "competing risks" that appear in medical research and industrial product development. We will teach you how to propose innovative models and acquire the ability to apply accurate statistical inference. We provide comprehensive guidance from the basics to practice, including the implmentation of R packages and online web applications. Copulas, Survival analysis, censoring
Kengo Kamatani
My teaching area covers Bayesian statistics and computational methods for statistical analysis. In particular, I deal with Markov chain Monte Carlo methods, sequential Monte Carlo methods, and Kalman filters. I will also discuss about related topics include Markov chains and their convergence theory, hidden Markov models and stochastic processes. The aim of my teaching is to provide students with the three basic skills of computation, mathematics and statistical applications. Bayesian Computation Bayesian Statistics, Markov chain, Monte Carlo methods
Satoshi Kuriki
Research on theory and application of mathematical statistics, multivariate analysis, and related mathematics. (1) Statistical inference theory and numerical solution in multivariate analysis, including continuous multivariate data analysis, contingency table analysis, and graphical model, etc. (2) Integral geometry and stochastic geometry approach to statistical distribution theory, including analysis of random fields, topological data analysis, and astronomical data analysis, etc. (3) Applied mathematics methods in statistical science, including algebraic methods, discrete mathematics and combinatorics. Special Topics in Statistical Inference and Mathematics 2 Mathematical statistics, Multivariate analysis, Geometric approach
Jiancang Zhuang
This course teaches the probability theory of point processes, including the concepts of random measures, Janossy density, Janossy measure, Campbell measure, moment measure, conditional intensity, Papangelou intensity, Palm intensity, etc.. In addition, it also gives an introduction on the techniques related to statistical inferences for random events in time and/or geographical space. In details, we focus on the issues of model construction, information recognition, model diagnostics, model selection, simulation, forecasting, forecast evaluation, etc.. Basic theory of Point Processes Point process, Statistical modelling and forecasting, Stochastic process
Yoshiyuki Ninomiya
In this course, we will discuss about irregular statistical models in which conventional statistical asymptotic theory does not hold and about estimation methods for the case in which standard maximum likelihood estimation is not appropriate. Specifically, we will treat (1) change-point models in which the likelihood cannot be differentiated and models with non-identifiability such as a signal model, a mixture model and a factor model, and (2) sparse estimations and semiparametric estimations using propensity scores. Irreggular Statistical Theory causal inference, model selection, sparse estimation
Hironori Fujisawa
This course focuses on statistical inference, statistical machine learning, and related data analyses. Topic: Robust Statistics. Divergence. Sparse Modeling. Graphical Modeling. Asymmetry Distribution. Model Selection. Mixed Effects Models. Missing Data Analysis. Multiple Testing. Anomaly Detection. Data: Medical Data. Industrial Data. Genome Data. Statistical Inference Statistical Inference, Statistical Machine Learning, Statistical Data Analysis.
Shuhei Mano
My research interest is in statistical and stochastic problems mainly around discrete random structures appear in various fields and related with combinatorics, computational algebra, special functions, and differential geometry. In particular, I mainly forcus on sampling algorithms and statistical inferences in graphical models including contingency tables, integer partitions, urn models, discrete random measures, random graphs, and interacting particle systems. I also keep interest in methods and plactices of data analysis in cutting-edge data-drien sciences. Stochastic Models Algebraic Statistics, Bayesian Statistics, Stochastic Models
Shogo Kato
This course discusses theory and application of mathematical statistics. Examples of the topics discussed in the course include: (1) statistics on non-Euclidean spaces, especially, directional statistics and statistical shape analysis; (2) parametric statistical models including probability distributions and regression models; and (3) copulas and related dependence measures. The goal of this course is to develop statistical methods, which are mathematically tractable and useful in practice, and consider their applications. Parametric Statistical Model Directional statistics, Distribution theory, Regression analysis
Ayaka Sakata
We study approximation methods for inference problems. In particular, mean field theory and approximation algorithms are key to our research activity. Through the researches on sparse estimation and Bayesian modeling, which can be regarded as problems of high-dimensional random systems, we study analytical and numerical methods for approximated inference quantifying its accuracy and credibility. Mean field theory for random system Mean-field theory, Sparse estimation, Approximate Bayesian inference
Takaaki Shimura
Probability theory is basic mathematics for statistical science. As statistical science develops, higher mathematics is required. I deal with several mathematical topics for statistical science: (1) Infinitely divisible distributions and processes. This distribution and process class is an important generalization of Gaussian and Poisson distributions and processes. (2) Extreme value theory and its applications. Extreme value or order statistics is interesting from both theoretical and practical points of view. These are used for mathematical modelling in financial engineering, insurance, natural disasters, risk control and so on. Special Topics in Statistical Inference and Mathematics 1 Lévy process, Extreme value theory, Regular variation
Mirai Tanaka
We study theory, numerical algorithms, and applications of mathematical optimization, mainly continuous optimization. Specific topics include (1) numerical algorithms and related theory for solving large-scale nonlinear optimization, conic optimization, and related problems, and (2) their applications to various fields including operations research, statistical sciences, etc. Topics in Computational Mathematics Continuous optimization, Mathematical optimization, Operations research
Ikuko Funatogawa
This course focus on the study of statistical models, such as linear mixed-effects models and their extensions, used in the longitudinal data analysis in which a response variable is measured repeatedly over time for multiple subjects. The course will also focus on the study of research designs, such as randomization, and on the study of statistical methods used in actual problems. Longitudinal Data Analysis Longitudinal Data Analysis, Linear Mixed Effects Models
Bruno Figueira Lourenço
I supervise students on topics related to continuous optimization, typically with a focus on the theoretical aspects but without losing sight of the practice. Here are a few examples: 1. Conic linear programming: algorithms, regularization techniques, ill-posedness and error bounds. 2. Algorithms and optimality conditions for nonlinear conic programming. 3. Mathematics related to continuous optimization: convex analysis, Jordan algebras and etc. 4. Nonsmooth optimization Convex Analysis and Conic Optimization Continuous Optimization, Conic Optimization, Nonsmooth Optimization
Masayuki Henmi
In this course, we mainly do research on modern statistical methods of biostatistics. More concretely, this includes missing-data analysis, statistical causal inference, semiparametric inference, meta analysis and so on. These are used in other areas such as social science, and it is expected that the students have interest in such aspect and will have a wider view through statistical methodology. Topics of Statistical Inference Semiparametric inference, Information geometry, Statistical theory
Daisuke Murakami
Methodology and application of spatial and spatio-temporal statistics. Recent main topics are as follows: (1)Gassian process and regression modeling focusing on spatial and temporal correlations; (2) fast and flexble modeling for a wide variety of spatial data; (3) Application to urban and regional problems related to economy, environment, disease, etc. Spatial statistics, spatio-temporal modeling, urban and environmental analysis
Daichi Mochihashi
Provide education and research supervision on statistical natural language processing and related statistical machine learning problems on discrete domain. This includes research on natural language processig, linguistics, robotics, educational statistics, music and audio processing, computational social science and institutional research. Machine Learning for Statistical Natural Language Processing Natural language processing, Bayesian statistics, Nonparametric Bayesian methods
Keisuke Yano
The main goal of this course is to provide theory and applications of mathematical statistics. Specifically, students will be taught (1) Theory of predictive densities (2) High-dimensional and infinite-dimensional statistics (3) Analysis of seismic and geodetic data. Evaluation of prediction-based methods such as Akaike Information Criterion and cross-validation is a fundamental tool in data analysis. One of the aims of this course is to understand the nature of data analysis methods through prediction and to construct new methods. High dimensional probability and statics Predictive density, High and infinite-dimensional statistics, Data analysis on seismic and geodetic data

(Assistant Professor) Akifumi Okuno / Kei Noba

Department of Interdisciplinary Statistical Mathematics

Specialty Courses Keywords
Genta Ueno
I give students guidance in their studies of methodologies and applications of data assimilation. Data assimilation is a method of time-series analysis for large-scale data set on the basis of a numerical model that simulates the time evolution of the states. The combination of the simulation model and the data set enables us to predict and estimate the states more accurately than with the data or the model alone. Keeping realistic applications in mind, I give guidance in methods of time-series analysis, statistical models, implementation with parallel computing, and methods for analyzing the resultant estimates. Special Course on Data Assimilation Data assimilation, Numerical simulation, Parallel computing
Koji Kanefuji
This course is intended to follow on from statistical inference by providing a more specical topics of the applied fields such as biomertics and environmetrics. The focus will be on developing a deeper theoretical understanding of some of the important statistical methods. (1)Statistical methods for survival data analysis (2)Analysis of longitudinal data Special Topics in Environmental Statistics Biometrics, Environmetrics, Lifetime Distribution
Yoshinori Kawasaki
Students will be guided to learn various time series models useful to time series econometrics. Possible topics are unit root tests, cointegration, vector autoregressive model, conditional heteroscedasticity, conditional duration and so on. Also, instructions will be given for statistical aspects of time series models with unobservable components in connection with pricing and prediction of financial assets. Special Topics in Time Series Analysis Non-stationary time series analysis, conditional heteroscedasticity, multiple time series
Shin'ya Nakano
My research interests are in analyses of spatio-temporal data and their applications to estimation and prediction of a system. In particular, the main subject is the methodology of data assimilation, which exploits our knowledge on system dynamics for the estimation. Analyses of spatio-temporal data based on the state space model and statistical modelling for data assimilation and analyses of various spatio-temoral data are also covered in the scope of my interest. Students who are interested in these research subjects are welcome. Statistical Computing data assimilation, state space model, statistical emulator
Hisashi Noma
Methodology and application of biostatistics. (1) Designs and analyses of clinical and epidemiological researches (2) Evidence synthesis methods (meta-analysis) (3) Prevention and analyses of missing data in medical studies (4) Analyses of large-scale genomic data, etc. Special Topics in Biostatistics Biostatistics, Applied statistics, Clinical epidemiology
Tomoko Matsui
The main goal of this course is to introduce the students to influential developments in modern machine learning for communication and multimedia information processing, namely kernel machine, hidden Markov models, graphical models and deep learning models. The following applications to real-world problems with big data will serve as examples. (1) speech and speaker recognition (2) multimedia data classification Multimedia Information Processing Machine learning, State space model, Kernel machine, Deep learning model
Kazuhiro Minami
This course covers information security, particularly privacy-preserving techniques for protecting sensitive information in big data. We focus on recent research on anonymization and statistical disclosure control and learn important elements of security research, such as adversary models, safety metrics, and design methodologies and evaluation methods of privacy-preserving algorithms.  Information Security Anonymization, Statistical Disclosure Control, Differential Privacy
Satoshi Yamashita
For stochastically occurring social risks, we will conduct research on grasping the amount of risk and establishing a risk response system and risk model evaluation method. The target fields for this year are (1) Credit risk quantification and its management method (2) Finance-related forecast models such as market fluctuations and their evaluation methods (3) Government survey Corporate evaluation using micro data (4) Analysis of databases with multiple and complex structures such as real estate data. Financial Statistics Credit risk, Market risk, Decision making under risk
Atsushi Yoshimoto
Teaching and advising focus on statistical and mathematical modeling for predicting and controlling natural and socio-economic resource change within the deterministic and stochastic frameworks. Through field survey, research study will be conducted on sustainable resource management as a socio-economic system along with risk evaluation and economic analysis. Applied Probability Discrete Optimization, Mathematical Programming, Resource Management
Jun Adachi
In this course, students will learn about modeling of molecular evolution and estimating of phylogenetic trees.
Specific content is as follows:
major transitions of evolution;
genomic evolution;
comparison of DNA sequences to calculate genetic distances;
phylogeny reconstruction by distance, likelihood and Bayesian methods;
test of the molecular clock and dating speciation events;
mechanisms of molecular evolution and the neutral theory.
Genomic Data Analysis
Shinsuke Koyama
Theories and applications of stochastic methods for modeling and analyzing complex phenomena observed in natural and social systems: 1) Fundamental theory of Stochastic processes; 2) Analytical methods such as SDEs, Master equations and Fokker-Planck equations; 3) State-Space models, Kalman filter, nonlinear filtering; 4) Bayesian modeling; 5) Stochastic modeling in science and engineering. Stochastic Modeling Stochastic processes, state-space models, Bayesian modeling
Kenichiro Shimatani
How to construct spatio-temporal models when spatio-temporal data are given? How to formulate system and observation models by mathematical equations and probability distributions? How to estimate unknown parameters and how to check the goodness-of-fit ? Primarily, spatio-temporal field data about plants and animals are analyzed. Spatial Statistics and Stochastic Geometry Point process, Ecology, Spatio-temporal model
Yumi Takizawa
Many of data and signals in nature world and social activity are diverse and complex. To create effective system for these analysis and computation, The basis of digital signal processing will be lectured. Next, the basis of information theory by C. Shannon will be lectured.
Furthermore, neural systems in brain of animal including human are studied as higher function and flexible system of small power consumption. the goal of these studies are to establish a novel methodology for computation, communication, and mesurement.
Special Topics in Modeling 1 Space-time analysis, Remote sensing systems, Radio communication systems
Yoosung Park
Practical methodologies related to social surveys on individuals and small groups in organizations or communities. (1) Surveys and statistical analyses on organizational behavior (2) Empirical research and applications for municipality residents survey (3) Experiments and techniques for improving mail survey methodology Survey Design Organizational Behavior Survey, Municipality Residents Survey, Mail Survey Methodology
Tadahiko Maeda
Students will be instructed on the process from data collection to analysis of survey data, using social surveys as a typical material. We will discuss the theory and practice of the sample survey as well as various issues related to the actual survey process. For comparison, data acquisition and analysis of other types of data than social surveys are also discussed. Data analysis methods will be examined, focusing on the practical issues of typical multivariate analysis methods applied to survey data. Topics in Sampling Theory Social Survey, Sampling Theory, Survey Data Analysis
Fumikazu Miwakeichi
I provide research guidance on spatio-temporal modeling for feature extraction and system structure estimation of complex feedback systems in natural science and engineering systems. Using mainly bio-signal data such as neurological data as an example, I will cover various methods related to practical spatio-temporal filtering, causal analysis, network estimation, signal separation and reconstruction, digital image processing, and visualization. Complex Systems Analysis Spatio-temporal analysis, Causal analysis, Neuroinformatics
Takao Murakami
This course focuses on privacy-preserving technologies and privacy metrics for data analysis and machine learning. It covers a wide range of topics related to privacy protection, from theory to practice. The main topics include (but are not limited to): (1) privacy-preserving data analysis and machine learning based on differential privacy, (2) privacy and utility analysis of privacy-preserving mechanisms, and (3) new privacy metrics to achieve high privacy and utility. privacy protection, privacy metrics, differential privacy

(Assistant Professor)Nobuo Shimizu

School of Statistical Thinking

(Assistant Professor)Ryota Yuasa