Committee |
Date Time |
Place |
Paper Title / Authors |
Abstract |
Paper # |
NC, IBISML, IPSJ-BIO, IPSJ-MPS [detail] |
2022-06-28 15:50 |
Okinawa |
(Primary: On-site, Secondary: Online) |
Causal Discovery in Discrete Data Using NML Code Length Based on MDL Principle Masatoshi Kobayashi, Nishimoto Hiroki, Shin Mastushima (Todai) NC2022-21 IBISML2022-21 |
Inference on the causal structure among random variables from only a finite number of observed data is one of the most i... [more] |
NC2022-21 IBISML2022-21 pp.149-155 |
IT, EMM |
2020-05-29 11:45 |
Online |
Online |
On MDL Estimation for Location Parameters of Contaminated Gaussian Distributions Kohei Miyamoto, Jun'ichi Takeuchi (Kyushu Univ.) IT2020-10 EMM2020-10 |
The MDL estimator is defined by the minimizer of total code length of a two part code.
The smaller the redundancy of t... [more] |
IT2020-10 EMM2020-10 pp.55-60 |
PRMU, IBISML, IPSJ-CVIM [detail] |
2018-09-21 10:00 |
Fukuoka |
|
A Study on MDL Learning of Mixture Families Kohei Miyamoto, Jun'ichi Takeuchi (Kyushu Univ.) PRMU2018-51 IBISML2018-28 |
MDL estimators based on two stage codes are studied.
The Barron and Cover's theory shows that redundancy of
a two stag... [more] |
PRMU2018-51 IBISML2018-28 pp.105-112 |
PRMU, IBISML, IPSJ-CVIM [detail] |
2017-09-15 13:00 |
Tokyo |
|
On MDL Learning of Gaussian Mixture Modlels Kohei Miyamoto, Masanori Kawakita, Jun'ichi Takeuchi (Kyushu Univ.) PRMU2017-47 IBISML2017-19 |
The final goal of this work is model sellection for gaussian mixture models(GMM) based on the minimum description length... [more] |
PRMU2017-47 IBISML2017-19 pp.59-66 |
IT |
2017-09-08 14:50 |
Yamaguchi |
Centcore Yamaguchi Hotel |
On Two Part Coding of Gaussian Mixture Models Kohei Miyamoto, Masanori Kawakita, Jun'ichi Takeuchi (Kyushu Univ.) IT2017-47 |
The final goal of this work is model sellection for gaussian mixture
models(GMM) based on the minimum description
leng... [more] |
IT2017-47 pp.49-54 |
SIS |
2016-06-09 15:00 |
Hokkaido |
Kushiro Tourism and Convention cent. |
Enumeration of coherent signals and independent signals based on MDL principle Masakiyo Suzuki, Yasunari Maeda (KIT) SIS2016-7 |
This paper shows an exact formulation of the MDL (Minimum Description Length) criterion for simultaneously enumerating t... [more] |
SIS2016-7 pp.35-40 |
IT |
2015-07-13 15:15 |
Tokyo |
Tokyo Institute of Technology |
Design and Analysis of MDL Estimators for Supervised Learning Jun'ichi Takeuchi, Masanori Kawakita (Kyushu Univ.) IT2015-26 |
Barron and Coveys theory
to evaluate risk bounds for the MDL estimators (1991)
is basically for unsupervised learning... [more] |
IT2015-26 pp.53-58 |
NC, IPSJ-BIO, IBISML, IPSJ-MPS (Joint) [detail] |
2015-06-23 17:00 |
Okinawa |
Okinawa Institute of Science and Technology |
Risk Bound of Lasso Based on MDL Theory Masanori Kawakita, Yushin Toyokihara, Jun'ichi Takeuchi (Kyushu Univ.) IBISML2015-16 |
We derive a risk bound of lasso in random design linear regression. Past
works of MDL principle revealed that penalized... [more] |
IBISML2015-16 pp.101-107 |
PRMU, IBISML, IPSJ-CVIM (Joint) [detail] |
2012-09-02 10:30 |
Tokyo |
|
Detecting Changes of Graph Partitioning Structures using Stochastic Decision Trees Shoichi Sato, Kenji Yamanishi (Univ. of Tokyo) PRMU2012-31 IBISML2012-14 |
We are concerned with the issue of estimating graph partitioning structures
from time series and tracking their changes... [more] |
PRMU2012-31 IBISML2012-14 pp.9-16 |
IBISML |
2012-06-19 - 2012-06-20 |
Kyoto |
Campus plaza Kyoto |
Detecting changes of graph partitioning structures Shoichi Sato, Kenji Yamanishi (Univ. of Tokyo) |
We are concerned with the issue of estimating graph partitioning structures
from time series and tracking their changes... [more] |
|
IBISML |
2012-06-20 14:50 |
Kyoto |
Campus plaza Kyoto |
An MDL-based Change-Detection Algorithm for Piecewise Stationay Memoryless Sources with Multi-dimensional Parameter Hiroki Kanazawa, Kenji Yamanishi (Univ. of Tokyo) IBISML2012-9 |
Kleinberg has proposed an algorithm for detecting bursts from a data sequence, which has turned out to be effective in t... [more] |
IBISML2012-9 pp.57-64 |
IBISML |
2011-11-10 15:45 |
Nara |
Nara Womens Univ. |
Dynamic model selection with resetting distributions Eiichi Sakurai (AIST), Kenji Yamanishi (The Univ. of Tokyo) IBISML2011-66 |
We are concerned with the issue of tracking changes of statistical models (e.g. the number of parameters, a discrete mod... [more] |
IBISML2011-66 pp.163-168 |
IBISML |
2011-06-20 16:15 |
Tokyo |
Takeda Hall |
Efficient Computation of Re-Normalized Maximum Likelihood Coding for Gaussian Mixtures with Its Applications to Optimal Clustering So Hirai, Kenji Yamanishi (Univ. of Tokyo) IBISML2011-5 |
We are concerned with the issue of efficient computation of re-normalized maximum likelihood (RNML) code-lengths for Gau... [more] |
IBISML2011-5 pp.29-35 |
IT |
2011-05-20 11:25 |
Osaka |
Osaka City University |
A Generalized the Chow-Liu Algorithm based on the MDL Principle Joe Suzuki (Osaka Univ.) IT2011-2 |
We extend the Chow-Liu algorithm for general random variables while
the previous versions only considered finite source... [more] |
IT2011-2 pp.7-12 |
ISEC, IT, WBS |
2011-03-04 09:50 |
Osaka |
Osaka University |
A Generalized MDL Criterion and its Application to Learning Bayesian Network Structure
-- When Both Discrete and Continuous Data are Present -- Joe Suzuki (Osaka Univ.) IT2010-106 ISEC2010-110 WBS2010-85 |
We propose the MDL principle without assuming the source to be discrete.
As an application, we consider estimation of B... [more] |
IT2010-106 ISEC2010-110 WBS2010-85 pp.249-254 |
IT |
2010-11-30 14:30 |
Nagano |
|
[Invited Talk]
Recent Advances in Information-Theoretic Learning Theory
-- Tracking Latent Dynamics -- Kenji Yamanishi (Univ. of Tokyo) IT2010-51 |
This paper addresses the issue of detecting changes of latent strucures behind data.The theory of probabilsitic models w... [more] |
IT2010-51 pp.1-8 |
IBISML |
2010-11-05 15:30 |
Tokyo |
IIS, Univ. of Tokyo |
[Poster Presentation]
Efficient Computation of Normalized Maximum Likelihood Coding for Gaussian Mixtures with Its Applications to Model Selection So Hirai, Kenji Yamanishi (Tokyo Univ.) IBISML2010-103 |
We are concerned with the issue of efficient computation of normalized maximum likelihood (NML) code-lengths for Gaussia... [more] |
IBISML2010-103 pp.327-333 |
IBISML |
2010-06-15 16:45 |
Tokyo |
Takeda Hall, Univ. Tokyo |
A linear time algorithm for sequential dynamic model selection Eiichi Sakurai, Kenji Yamanishi (Univ. of Tokyo.) IBISML2010-25 |
This paper addresses the issue of dynamic model selection (DMS), in which the goal is to select an optimal model sequenc... [more] |
IBISML2010-25 pp.175-180 |
IBISML |
2010-06-15 17:15 |
Tokyo |
Takeda Hall, Univ. Tokyo |
Graph Clustering based on Normalized Maximum Likelihood Coding So Hirai, Ryota Tomioka, Kenji Yamanishi (Univ. of Tokyo) IBISML2010-27 |
This paper addresses the issue of graph clustering, i.e., assigning nodes for a given graph into a number of clusters, i... [more] |
IBISML2010-27 pp.189-195 |
NC |
2006-03-15 14:30 |
Tokyo |
Tamagawa University |
Proposal of a Bayesian Network Structure Learning with Modified MDL Mori Kurokawa, Akito Sakurai (Keio Univ.), Chihiro Ono (KDDI R&D Labs.), Yoichi Motomura, Hideki Asoh (AIST) |
Minimum Description Length(MDL) criteria is generally used as model selection criteria for model construction in Bayesia... [more] |
NC2005-122 pp.97-101 |