MODAL - 2019 - Annual activity report

MODAL

MODAL - 2019

Project-Team Modal

Team, Visitors, External Collaborators

Overall Objectives

Context
Goals

Research Program

Research axis 1: Unsupervised learning
Research axis 2: Performance assessment
Research axis 3: Functional data
Research axis 4: Applications motivating research

Application Domains

Economic world
Biology

Highlights of the Year

New Software and Platforms

MixtComp
BlockCluster
CloHe
PACBayesianNMF
pycobra
STK++
rtkore
MixAll
simerge
MixtComp.V4
MASSICCC
Platforms

New Results

Axis 1: Data Units Selection in Statistics
Axis 1: Model-Based Co-clustering for Ordinal Data of different dimensions
Axis 1: Model-based co-clustering for mixed type data
Axis 1: Relaxing the Identically Distributed Assumption in Gaussian Co-Clustering for High Dimensional Data
Axis 1: Gaussian-based visualization of Gaussian and non-Gaussian model-based clustering
Axis 1: Co-clustering: A versatile way to perform clustering
Axis 1: Dealing with missing data in model-based clustering through a MNAR model
Axis 1: Organized Co-Clustering for textual data synthesis
Axis 1: Model-Based Co-clustering with Co-variables
Axis 1: Linking canonical and spectral clustering
Axis 1: Predictive clustering
Axis 1: Ranking and synchronization from pairwise measurements via SVD
Axis 1: SPONGE: A generalized eigenproblem for clustering signed networks
Axis 2: Multi-kernel unmixing and super-resolution using the Modified Matrix Pencil method
Axis 2: Provably robust estimation of modulo 1 samples of a smooth function with applications to phase unwrapping
Axis 2: Learning general sparse additive models from point queries in high dimensions
Axis 2: Sparse non-negative super-resolution - simplified and stabilized
Axis 2: Pseudo-Bayesian learning with kernel Fourier transform as prior
Axis 2: PAC-Bayesian binary activated deep neural networks
Axis 2: Improved PAC-Bayesian Bounds for Linear Regression
Axis 2: Multiview Boosting by controlling the diversity and the accuracy of view-specific voters
Axis 2: PAC-Bayes and Domain Adaptation
Axis 2: Interpreting Neural Networks as Majority Votes through the PAC-Bayesian Theory
Axis 2: Still no free lunches: the price to pay for tighter PAC-Bayes bounds
Axis 2: PAC-Bayesian Contrastive Unsupervised Representation Learning
Axis 2: Sequential Learning of Principal Curves: Summarizing Data Streams on the Fly
Axis 2: PAC-Bayes Un-Expected Bernstein Inequality
Axis 2: Attributing and Referencing (Research) Software: Best Practices and Outlook from Inria
Axis 2: Revisiting clustering as matrix factorisation on the Stiefel manifold
Axis 2: A Primer on PAC-Bayesian Learning
Axis 2: Perturbed Model Validation: A New Framework to Validate Model Relevance
Axis 2: Decentralized learning with budgeted network load using Gaussian copulas and classifier ensembles
Axis 2: Online k-means Clustering
Axis 2: Non-linear aggregation of filters to improve image denoising
Axis 2: Multiple change-points detection with reproducing kernels
Axis 2: Analysis of early stopping rules based on discrepancy principle
Axis 3: Short-term air temperature forecasting using Nonparametric Functional Data Analysis and SARMA models
Axis 3: Mathematical Modeling and Study of Random or Deterministic Phenomena
Axis 3: Categorical functional data analysis
Axis 4: Proteomic signature of early death in heart failure patients
Axis 4: Statistical analysis of high-throughput proteomic data
Axis 4: Linking different kinds of Omics data through a model-based clustering approach
Axis 4: Real-time Audio Sources Classification
Axis 4: Matching of descriptors evolving over time
Axis 4: Supervised multivariate discretization and levels merging for logistic regression
Axis 4: MASSICCC Platform for SaaS Software Availability
Axis 4: Domain adaptation from a pre-trained source model
Axis 4: Reject Inference Methods in Credit Scoring: a rational review
Other: Projection Under Pairwise Control

Bilateral Contracts and Grants with Industry

Bilateral Contracts with Industry
Bilateral Grants with Industry

Partnerships and Cooperations

Regional Initiatives
National Initiatives
European Initiatives
International Initiatives
International Research Visitors

Dissemination

Promoting Scientific Activities
Teaching - Supervision - Juries
Popularization

Bibliography

Major publications
Publications of the year

Previous |

Home | Next next

Section: New Results

Axis 2: A Primer on PAC-Bayesian Learning

Participant: Benjamin Guedj

This survey on PAC-Bayesian learning has been the backbone to a successful proposal for an ICML 2019 plenary tutorial.

Generalised Bayesian learning algorithms are increasingly popular in machine learning, due to their PAC generalisation properties and flexibility. The present paper aims at providing a self-contained survey on the resulting PAC-Bayes framework and some of its main theoretical and algorithmic developments.

This work has been published in the proceedings of the French Mathematical Society. Published as [66].

Previous |

Home | Next next