no code implementations • 16 Apr 2024 • Eduardo Fernandes Montesuma, Fred Ngolè Mboula, Antoine Souloumiac
Based on this novel algorithm, we propose two new strategies for MSDA: GMM-WBT and GMM-DaDiL.
no code implementations • 18 Mar 2024 • Eduardo Fernandes Montesuma, Fred Maurice Ngolè Mboula, Antoine Souloumiac
The optimal transport solution gives us a matching between source and target domain mixture components.
no code implementations • 14 Sep 2023 • Eduardo Fernandes Montesuma, Fred Ngolè Mboula, Antoine Souloumiac
To solve it, we adapt previous works in the MSDA literature, such as Wasserstein Barycenter Transport and Dataset Dictionary Learning, as well as DD method Distribution Matching.
no code implementations • 14 Sep 2023 • Fabiola Espinoza Castellon, Eduardo Fernandes Montesuma, Fred Ngolè Mboula, Aurélien Mayoue, Antoine Souloumiac, Cédric Gouy-Pailler
The proposed framework, FedDaDiL, tackles the resulting challenge through dictionary learning of empirical distributions.
no code implementations • 22 Aug 2023 • Eduardo Fernandes Montesuma, Michela Mulas, Fred Ngolè Mboula, Francesco Corona, Antoine Souloumiac
Nonetheless, these models are sensible to changes in the data distributions, which may be caused by changes in the monitored process, such as changes in the mode of operation.
Multi-Source Unsupervised Domain Adaptation Unsupervised Domain Adaptation
1 code implementation • 27 Jul 2023 • Eduardo Fernandes Montesuma, Fred Ngolè Mboula, Antoine Souloumiac
Based on our dictionary, we propose two novel methods for MSDA: DaDil-R, based on the reconstruction of labeled samples in the target domain, and DaDiL-E, based on the ensembling of classifiers learned on atom distributions.
no code implementations • 28 Jun 2023 • Eduardo Fernandes Montesuma, Fred Ngolè Mboula, Antoine Souloumiac
Recently, Optimal Transport has been proposed as a probabilistic framework in Machine Learning for comparing and manipulating probability distributions.
no code implementations • 5 Oct 2022 • Maciej Śliwowski, Matthieu Martin, Antoine Souloumiac, Pierre Blanchart, Tetiana Aksenova
The performance gap is reduced with bigger datasets, but considering the increased computational load, end-to-end training may not be profitable for this application.
no code implementations • 8 Sep 2022 • Maciej Śliwowski, Matthieu Martin, Antoine Souloumiac, Pierre Blanchart, Tetiana Aksenova
In this study, we investigated the impact of long-term recordings on motor imagery decoding from two main perspectives: model requirements regarding dataset size and potential for patient adaptation.
no code implementations • 5 Oct 2021 • Maciej Śliwowski, Matthieu Martin, Antoine Souloumiac, Pierre Blanchart, Tetiana Aksenova
These models have a limited representational capacity and may fail to capture the relationship between ECoG signal and continuous hand movements.
no code implementations • 12 Feb 2018 • Anne Morvan, Antoine Souloumiac, Krzysztof Choromanski, Cédric Gouy-Pailler, Jamal Atif
The aim of this paper is to endow the well-known family of hypercubic quantization hashing methods with theoretical guarantees.
no code implementations • 22 May 2017 • Anne Morvan, Antoine Souloumiac, Cédric Gouy-Pailler, Jamal Atif
We demonstrate the quality of our binary sketches through experiments on real data for the nearest neighbors search task in the online setting.