Releted seminars

日時: 2007年12月6日 (木) 13:30 - 16:00

場 所: 統計数理研究所 講堂


13:30 - 14:45 Friedrich Leisch (Institute of Statistics, Ludwig-Maximilians-Univ., Germany)

Finite Mixtures of Generalized Linear Regression Models

資料 (PDF)

14:45 - 16:00 Luke Tierney (Dept. of Statistics and Actuarial Science, Univ. of Iowa, USA)

MRI Tissue Classification Using Bayesian Hidden Markov Normal Mixture Models

資料 (PDF)


日時:2007年12月7日 (金) 10:30 - 18:00

    2007年12月8日 (土) 9:30 - 17:20

場所:統計数理研究所 講堂


10:30 - 11:30

Luke Tierney (University of Iowa)

Simple Parallel Statistical Computing in R

資料 (PDF)

11:45 - 13:30 Lunch

13:30 - 14:30

Friedrich Leisch (Ludwig-Maximilians-Universität)

R Behind the Scenes: Using S the (un)usual Way

資料 (PDF)

14:45 - 15:45

Luke Tierney (University of Iowa)

Code Analysis and Parallelizing Vector Operations in R

資料 (PDF)

16:00 - 17:00

Friedrich Leisch (Ludwig-Maximilians-Universität)

Flexible Implementations of Cluster Analysis and Mixture

資料 (PDF)

18:00 - Dinner

Each talk has one hour length and followed by about 10 minutes discussion.


9:40 - 10:30

牧山文彦 (データキューブ) 「R+M=?-Rとオブジェクトデータ ベースの新次元-」

資料 (PDF)

10:30 - 11:20

樋口千洋 (大日本住友製薬) 「Bioconductorの概要とその利用方法」

資料 (PDF)

11:20 - 12:10

鈴木了太 (ef-prime) 「ビジネスデータ分析の現場から」

資料 (PDF)

12:10 - 13:20 昼休み

13:20 - 14:10

舟尾暢男 (武田薬品工業) 「今日からあなたも統計ソフト開発者!~ R Commander の概要,改造,数式処理機能の実装例 ~」

資料 (PDF)

14:10 - 15:00

石田基広 (徳島大学総合科学部) 「Rと自然言語研究」

資料 (PDF)

15:00 - 15:20 休憩

15:20 - 16:10

谷村晋 (立命館アジア太平洋大学) 「疫学におけるRの活用」

資料 (PDF)

16:10 - 16:50

小笠原理 (遺伝研DDBJ) 「R Graphical Manualのつかいかた」

資料 (PDF)

16:50 - 17:20

山本寛 (NTT西日本),福崎昭伸 (NTT西日本),中野純司 (統計数理研究所) 「グリッドコンピューティングとR」

資料 (PDF)


Title: Simple Parallel Statistical Computing in R.

Speaker: Luke Tierney

This talk presents a framework for the R statistical computing.Language that provides a simple yet powerful parallel programmingi nterface to a computational cluster. The interface allows the development of R functions that distribute independent computations across the nodes of the cluster and thus can obtain significant speed-ups at little additional development cost. The basic framework will be introduced and illustrated using several examples. Joint work with A. J. Rossini and Na Li

Title: R Behind the Scenes: Using S the (un)usual Way.

Speaker: Friedrich Leisch

Most users know R as a statistical computing environments presenting them a prompt or minimalistic GUI for data analysis. The user enters commands and R responds with figures, tables, fitted models, etc. However, behind the prompt R is first of all an interpreter for a full-featured programming language named S. The prompt is only one way of utilizing that language, and numerous other ways have been developed over the last years. This presentation will give an overview of software solutions using R "indirectly": embedding R in other applications like spreadsheets, dynamic statistical documents combining text and code, using R as a scripting language, or as a webpage plugin offering a wide range of services from simple examples for teaching to complete data analyses over the Internet. For statisticians this often gives a simple way of making our methods accessible to a much wider audience.

Title: Code Analysis and Parallelizing Vector Operations in R.

Speaker: Luke Tierney

This talk will present some current work on two seemingly unrelated areas. The first is the development of code analysis tools to help identify possible errors in R code. Current versions of these tools have been useful in finding bugs in R's code as well as code in packages submitted to CRAN and are now enabled by default in package checking. The second area is the development of mechanisms to allow R's internal vectorized operations to take advantage of multiple processors. These two areas are related through their connections to ongoing efforts to develop a byte code compiler for R.

Title: Flexible Implementations of Cluster Analysis and Mixture.

Speaker: Friedrich Leisch
This talk is not only about cluster analysis and mixture models inparticular, but rather takes those as an example to show design priniciples for open and extensible programming practices (and hence the title may be "Design Principles for Open and Extensible R Packages"). That would close the circle to the more theoretical talk on mixture modelling on Thursday.

Title: MRI Tissue Classification Using Bayesian Hidden Markov Normal Mixture Models.

Speaker: Luke Tierney

This talk discusses the problem of identifying the major types of brain tissue, gray matter, white matter, and cerebrospinal fluid,based on magnetic resonance images (MRI). The measured intensities in the MR image are modeled as a normal mixture with a hidden Markov model accounting for the spatial similarity among nearby regions. A challenging issue, known as the partial volume effect, is that due to limited resolution of the images individual regions for which single measurements are available often contain more than one tissue type. This is addressed by modeling the data at a higher resolutionand viewing the observed measurements as aggregates of latent lower resolution measurements. Combined with an appropriate spatial prior distribution this produces very good results in both artificial and real test cases. Several Markov chain Monte Carlo approaches have been
explored to identify strategies effectively and produce good performance.
Joint work with Dai Feng.