Project-Team:MODAL

Project-Team Modal

Personnel

Overall Objectives

Research Program

Generative model design
Data visualization

Application Domains

Economic world
Biology

Highlights of the Year

New Software and Platforms

MixtComp
BlockCluster
CloHe
PACBayesianNMF
pycobra
STK++
rtkore
Platforms

New Results

An Oracle Inequality for Quasi-Bayesian Non-Negative Matrix Factorization
Simpler PAC-Bayesian Bounds for Hostile Data
Highlight 1 High-dimensional Adaptive Ranking with PAC-Bayesian Bounds
Online Adaptive Clustering
Study of Transcriptional Regulation
Functional Binary Linear Models for Stratified Samples
Mixture Model for Mixed Kind of Data
Data Units Selection in Statistics
Trade-off Between Computation Time and Accuracy
Projection Under Pairwise Control
Real-time Audio Sources Classification
Model-Based Co-clustering for Ordinal Data
Model-Based Co-clustering for Ordinal Data of different dimensions
MASSICCC Platform for SaaS Software Availability
Model-Based Co-Clustering of Multivariate Functional Data
Reject Inference Methods in Credit Scoring: A Rational Review
Survival Analysis with Complex Covariates: A Model-based Clustering Preprocessing Step
Dealing with Missing Data Through Mixture Models
Review on Mixture Modeling and High-dimensional Clustering
Dealing with Missing Not at Random Values in Model-based Clustering
Dealing with Several Cluster Variables

Bilateral Contracts and Grants with Industry

Florimond Desprez
Arcelor-Mittal
Alstom
Vallourec
Running Care

Partnerships and Cooperations

Regional Initiatives
National Initiatives
International Initiatives
International Research Visitors

Dissemination

Promoting Scientific Activities
Teaching - Supervision - Juries
Popularization

Bibliography

Major publications
Publications of the year

Inria | Raweb 2017 | Presentation of the Project-Team MODAL


	PDF	e-Pub

Previous |

Home | Next next

Section: New Results

Online Adaptive Clustering

Participant : Benjamin Guedj.

The quasi-Bayesian perspective has been extended to online adaptive clustering. Data streams are clustered dynamically with a quasi-Bayesian-flavored predictor, and we have proven minimax regret bounds. An efficient MCMC-based implementation is proposed.

Previous |

Home | Next next