Florent Bouchard

Centre National de la Recherche Scientifique · France

Unclaimed · AI-structured profileORCID OpenAlex

CNRS researcher applying random matrix theory, tensor decomposition and Riemannian learning to EEG and fMRI signals

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-17) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.

Research profile

Output—

Output data is insufficient: no normalized productivity or complete output metrics are available beyond raw yearly counts. This axis would measure publication volume and consistency, for which only per-year work counts (2016-2026) are currently recorded.

Impact—

Impact cannot be scored yet: detailed citation data are missing, with only an h-index and i10-index of 8 each on record. This axis would measure citation influence and how widely the researcher's methods are adopted by follow-up work.

Focus100/100

The focus score is 100/100, driven by an exceptionally concentrated field distribution: the field distribution lists a single field — machine learning for brain-computer interfaces (federated learning, Riemannian deep learning) — with full weight. The five leading topics (random matrices 12, blind source separation 11, advanced statistical methods 11, EEG/BCI 10, tensor decomposition 10 works) form one coherent methods-to-neuroimaging pipeline rather than scattered interests.

Collaboration39/100

The collaboration score is 39/100, reflecting a deliberately compact network: only 6 distinct co-authors are recorded across 24 works in the last three years, with recurring partners Thibault Pautrel, Ammar Mian, Guillaume Ginolhac, Ce Ju, Antoine Collas and Bertrand Thirion. In both reviewed 2026 papers the researcher sits in a middle author position, consistent with contributing inside small method-driven teams rather than large consortia.

Activity94/100

The activity score is 94/100, backed by an unbroken publication record from 2016 to 2026 and a clear acceleration: 2 works in 2016 rising to 11 in 2024 and 9 already recorded in 2026, with 24 works in the last three years. Output is uneven — 2019 and 2022 dip to 2 works each — but the recent trajectory shows sustained, intensifying productivity.

Reproducibility55/100

The reproducibility score is 55/100, computed from a single reviewed paper (reproMean 55 over 1 paper), so it should be read as provisional. That paper — personalised federated learning for Riemannian and Euclidean EEG decoding — carries a medium reproducibility readiness, and its Evolve verdict notes that one comparison against centralised training lacks statistical support, a caveat for exact replication.

Research summary

Florent Bouchard is a researcher at the French National Centre for Scientific Research (CNRS) whose work sits at the intersection of statistical signal processing, random matrix theory, and machine learning for neuroimaging. His OpenAlex topic profile is led by Random Matrices and Applications (12 works), Blind Source Separation Techniques (11), Advanced Statistical Methods and Models (11), EEG and Brain-Computer Interfaces (10), and Tensor decomposition and applications (10), indicating a methods core steadily redirected toward brain-data applications. His recorded metrics show an h-index and i10-index of 8 each, with works logged every year from 2016 to 2026, a peak of 11 works in 2024, and 9 already recorded in 2026; 24 works fall within the last three years, supporting an activity score of 94/100. The Evolve-reviewed 2026 paper on personalised federated learning for Riemannian and Euclidean EEG decoding shows his current focus on federated, Riemannian deep learning for brain-computer interfaces, pursued within a compact co-author network (6 distinct recent co-authors) including Thibault Pautrel, Ammar Mian, Guillaume Ginolhac, Ce Ju, Antoine Collas and Bertrand Thirion. A focus score of 100/100 reflects a single concentrated field: machine learning for brain-computer interfaces (federated learning, Riemannian deep learning).

Collaboration pattern

Collaboration is compact and stable: 6 distinct co-authors across 24 works in the last three years (collaboration dimension 39/100). The recurring circle — Thibault Pautrel, Ammar Mian, Guillaume Ginolhac, Ce Ju, Antoine Collas and Bertrand Thirion — spans signal-processing and machine-learning groups, and in both reviewed 2026 papers the researcher sits in a middle author position rather than first or corresponding, consistent with a specialist contributor inside small method-driven teams.

Research keywords

random matrix theoryblind source separationtensor decompositionRiemannian geometrystatistical signal processingEEG decodingbrain-computer interfacesfederated learningpersonalised federated learningRiemannian deep learningSPD matrix learningresting-state fMRI connectome prediction

Highlights

Personalised federated learning lifts Riemannian EEG decoding
Evolve's review reports that head-personalised federated learning applied over federated SPDNet outperforms both standard federated learning and centralised training on three EEG datasets, though one centralised win lacks statistical support; reproducibility readiness is rated medium.
One open question: statistical support for a centralised comparison
The same Evolve review flags that one of the reported wins over centralised training lacks statistical support — a concrete point the authors, and anyone replicating the benchmark, should tighten in follow-up work on personalised federated EEG decoding.
Benchmarking external generalization of SPD matrix learning
A 2026 arXiv study benchmarking how SPD matrix learning methods generalise externally for resting-state fMRI connectome prediction; it is in the Evolve catalog but has not yet been reviewed, so no verdict, keywords or reproducibility rating are available.

Timeline

2016Earliest recorded output: 2 works published in 2016, opening an unbroken run of annual publications through 2026.
2024Peak yearly output: 11 works recorded in 2024, the highest annual count in the 2016-2026 bibliometric record.
2026Personalised federated learning for Riemannian and Euclidean EEG decoding appears on arXiv; Evolve's review rates its reproducibility readiness as medium.
2026Benchmarking study on external generalization of SPD matrix learning for resting-state fMRI connectome prediction enters the Evolve catalog, not yet reviewed.

Papers in the Evolve catalog

2026-09-24
Personalised federated learning for Riemannian and Euclidean EEG decoding

Frequent co-authors

Thibault Pautrel×1 Ammar Mian×1 Guillaume Ginolhac×1 Ce Ju×1 Antoine Collas×1 Bertrand Thirion×1

About this profile

This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.

Bibliometrics are computed from OpenAlex records only; the output and impact dimensions lack sufficient underlying data and are not scored.
Only one catalog paper has a full Evolve review, so the reproducibility score (55) rests on a single reviewed paper.
The second catalog paper (SPD matrix learning benchmark) has no verdict, keywords or reproducibility rating yet.
Counts for 2026 cover a partial year, and this profile is name-resolved from public bibliographic data (ORCID 0000-0003-3003-7317).