Zobrazeno 1 - 10
of 15
pro vyhledávání: '"Lafon, Marc"'
Fully connected Graph Transformers (GT) have rapidly become prominent in the static graph community as an alternative to Message-Passing models, which suffer from a lack of expressivity, oversquashing, and under-reaching. However, in a dynamic contex
Externí odkaz:
http://arxiv.org/abs/2409.17986
Prompt learning has been widely adopted to efficiently adapt vision-language models (VLMs), e.g. CLIP, for few-shot image classification. Despite their success, most prompt learning methods trade-off between classification accuracy and robustness, e.
Externí odkaz:
http://arxiv.org/abs/2407.01400
Autor:
Lafon, Marc, Thomas, Alexandre
Combining empirical risk minimization with capacity control is a classical strategy in machine learning when trying to control the generalization gap and avoid overfitting, as the model class capacity gets larger. Yet, in modern deep learning practic
Externí odkaz:
http://arxiv.org/abs/2403.10459
In this work, we study the out-of-distribution (OOD) detection problem through the use of the feature space of a pre-trained deep classifier. We show that learning the density of in-distribution (ID) features with an energy-based models (EBM) leads t
Externí odkaz:
http://arxiv.org/abs/2403.10403
Publikováno v:
International Conference on Machine Learning, ICML 2023, 23-29 July 2023, Honolulu, Hawaii, USA
Out-of-distribution (OOD) detection is a critical requirement for the deployment of deep neural networks. This paper introduces the HEAT model, a new post-hoc OOD detection method estimating the density of in-distribution (ID) samples using hybrid en
Externí odkaz:
http://arxiv.org/abs/2305.16966
Autor:
Lafon, Marc
Des mesures de l'activité du cerveau humain peuvent être obtenues par diverses techniques d'imagerie neuro-fonctionnelle. Dans le domaine des neurosciences, les études menées grâce à ces nouvelles techniques sont appelées études d'activation.
Externí odkaz:
http://tel.archives-ouvertes.fr/tel-00010176
http://tel.archives-ouvertes.fr/docs/00/04/84/24/PDF/tel-00010176.pdf
http://tel.archives-ouvertes.fr/docs/00/04/84/24/PDF/tel-00010176.pdf
Autor:
Lafon, Marc.
Thèse--Faculté de théologie protestante de Montauban.
"Ouvrages consultés": 2d prelim. leaf.
"Ouvrages consultés": 2d prelim. leaf.
Externí odkaz:
http://catalog.hathitrust.org/api/volumes/oclc/29320350.html
Publikováno v:
ICML 2021 Workshop on Uncertainty and Robustness in Deep Learning
ICML 2021 Workshop on Uncertainty and Robustness in Deep Learning, Sep 2021, Virtual, Austria
ICML 2021 Workshop on Uncertainty and Robustness in Deep Learning, Sep 2021, Virtual, Austria
International audience; In this paper, we tackle the challenge of jointly quantifying in-distribution and out-of-distribution (OOD) uncertainties. We introduce KLoS, a KL-divergence measure defined on the classprobability simplex. By leveraging the s
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=dedup_wf_001::fb89fd0074b3cebdea17d7dfe24eea8b
https://hal-cnam.archives-ouvertes.fr/hal-03347628
https://hal-cnam.archives-ouvertes.fr/hal-03347628
Autor:
Pastor, Josette, Lafon, Marc, Travé-Massuyès, Louise, Démonet, Jean-Francois, Doyon, Bernard, Celsis, Pierre
Publikováno v:
Biological Cybernetics. 2000, Vol. 82 Issue 1, p49. 11p.
Autor:
Lafon, Marc
Publikováno v:
Modélisation et simulation. Université Paul Sabatier-Toulouse III, 2000. Français. ⟨NNT : ⟩
In the domain of neuroscience, activation studies use functional brain imaging techniques in order to provide measures of human brain activity during the performance of cognitive tasks. These studies highlight the fact that large-scale networks of ce
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=od______1398::12e8bdbdfaaf095f6094c55af82c4ef7
https://theses.hal.science/tel-00010176
https://theses.hal.science/tel-00010176