scispace - formally typeset
Search or ask a question
Institution

University of Grenoble

EducationSaint-Martin-d'Hères, France
About: University of Grenoble is a education organization based out in Saint-Martin-d'Hères, France. It is known for research contribution in the topics: Population & Large Hadron Collider. The organization has 25658 authors who have published 45143 publications receiving 909760 citations.


Papers
More filters
Journal ArticleDOI
TL;DR: A baseline solution to the aforementioned difficulty by developing a general multimodal deep learning (MDL) framework that is not only limited to pixel-wise classification tasks but also applicable to spatial information modeling with convolutional neural networks (CNNs).
Abstract: Classification and identification of the materials lying over or beneath the Earth's surface have long been a fundamental but challenging research topic in geoscience and remote sensing (RS) and have garnered a growing concern owing to the recent advancements of deep learning techniques. Although deep networks have been successfully applied in single-modality-dominated classification tasks, yet their performance inevitably meets the bottleneck in complex scenes that need to be finely classified, due to the limitation of information diversity. In this work, we provide a baseline solution to the aforementioned difficulty by developing a general multimodal deep learning (MDL) framework. In particular, we also investigate a special case of multi-modality learning (MML) -- cross-modality learning (CML) that exists widely in RS image classification applications. By focusing on "what", "where", and "how" to fuse, we show different fusion strategies as well as how to train deep networks and build the network architecture. Specifically, five fusion architectures are introduced and developed, further being unified in our MDL framework. More significantly, our framework is not only limited to pixel-wise classification tasks but also applicable to spatial information modeling with convolutional neural networks (CNNs). To validate the effectiveness and superiority of the MDL framework, extensive experiments related to the settings of MML and CML are conducted on two different multimodal RS datasets. Furthermore, the codes and datasets will be available at this https URL, contributing to the RS community.

582 citations

Journal ArticleDOI
TL;DR: In this paper, the authors present a model applicable to ultrasound contrast agent bubbles that takes into account the physical properties of a lipid monolayer coating on a gas microbubble, including buckling radius, the compressibility of the shell, and a break-up shell tension.
Abstract: We present a model applicable to ultrasound contrast agent bubbles that takes into account the physical properties of a lipid monolayer coating on a gas microbubble Three parameters describe the properties of the shell: a buckling radius, the compressibility of the shell, and a break-up shell tension The model presents an original non-linear behavior at large amplitude oscillations, termed compression-only, induced by the buckling of the lipid monolayer This prediction is validated by experimental recordings with the high-speed camera Brandaris 128, operated at several millions of frames per second The effect of aging, or the resultant of repeated acoustic pressure pulses on bubbles, is predicted by the model It corrects a flaw in the shell elasticity term previously used in the dynamical equation for coated bubbles The break-up is modeled by a critical shell tension above which gas is directly exposed to water

579 citations

Journal ArticleDOI
TL;DR: In this paper, a study was conducted to correlate the occurrence and severity of liver steatosis with hepatitis C virus type, level and sequence of the core-encoding region.

572 citations

Journal ArticleDOI
TL;DR: A new minibatch GCN is developed that is capable of inferring out-of-sample data without retraining networks and improving classification performance, and three fusion strategies are explored: additive fusion, elementwise multiplicative fusion, and concatenation fusion to measure the obtained performance gain.
Abstract: Convolutional neural networks (CNNs) have been attracting increasing attention in hyperspectral (HS) image classification due to their ability to capture spatial–spectral feature representations. Nevertheless, their ability in modeling relations between the samples remains limited. Beyond the limitations of grid sampling, graph convolutional networks (GCNs) have been recently proposed and successfully applied in irregular (or nongrid) data representation and analysis. In this article, we thoroughly investigate CNNs and GCNs (qualitatively and quantitatively) in terms of HS image classification. Due to the construction of the adjacency matrix on all the data, traditional GCNs usually suffer from a huge computational cost, particularly in large-scale remote sensing (RS) problems. To this end, we develop a new minibatch GCN (called miniGCN hereinafter), which allows to train large-scale GCNs in a minibatch fashion. More significantly, our miniGCN is capable of inferring out-of-sample data without retraining networks and improving classification performance. Furthermore, as CNNs and GCNs can extract different types of HS features, an intuitive solution to break the performance bottleneck of a single model is to fuse them. Since miniGCNs can perform batchwise network training (enabling the combination of CNNs and GCNs), we explore three fusion strategies: additive fusion, elementwise multiplicative fusion, and concatenation fusion to measure the obtained performance gain. Extensive experiments, conducted on three HS data sets, demonstrate the advantages of miniGCNs over GCNs and the superiority of the tested fusion strategies with regard to the single CNN or GCN models. The codes of this work will be available at https://github.com/danfenghong/IEEE_TGRS_GCN for the sake of reproducibility.

560 citations

Journal ArticleDOI
31 Oct 2018-Nature
TL;DR: In this article, an analysis of the kinematics, chemistry, age and spatial distribution of stars that are mainly linked to two major Galactic components: the thick disk and the stellar halo.
Abstract: The assembly of our Galaxy can be reconstructed using the motions and chemistry of individual stars1,2. Chemo-dynamical studies of the stellar halo near the Sun have indicated the presence of multiple components3, such as streams4 and clumps5, as well as correlations between the stars’ chemical abundances and orbital parameters6–8. Recently, analyses of two large stellar surveys9,10 revealed the presence of a well populated elemental abundance sequence7,11, two distinct sequences in the colour–magnitude diagram12 and a prominent, slightly retrograde kinematic structure13,14 in the halo near the Sun, which may trace an important accretion event experienced by the Galaxy15. However, the link between these observations and their implications for Galactic history is not well understood. Here we report an analysis of the kinematics, chemistry, age and spatial distribution of stars that are mainly linked to two major Galactic components: the thick disk and the stellar halo. We demonstrate that the inner halo is dominated by debris from an object that at infall was slightly more massive than the Small Magellanic Cloud, and which we refer to as Gaia–Enceladus. The stars that originate in Gaia–Enceladus cover nearly the full sky, and their motions reveal the presence of streams and slightly retrograde and elongated trajectories. With an estimated mass ratio of four to one, the merger of the Milky Way with Gaia–Enceladus must have led to the dynamical heating of the precursor of the Galactic thick disk, thus contributing to the formation of this component approximately ten billion years ago. These findings are in line with the results of galaxy formation simulations, which predict that the inner stellar halo should be dominated by debris from only a few massive progenitors2,16.

558 citations


Authors

Showing all 25961 results

NameH-indexPapersCitations
Dieter Lutz13967167414
Marcella Bona137139192162
Nicolas Berger137158196529
Cordelia Schmid135464103925
J. F. Macías-Pérez13448694715
Marina Cobal132107885437
Lydia Roos132128489435
Tetiana Hryn'ova131105984260
Johann Collot131101882865
Remi Lafaye131101283281
Jan Stark131118687025
Sabine Crépé-Renaudin129114282741
Isabelle Wingerter-Seez12993079689
James Alexander12988675096
Jessica Levêque129100670208
Network Information
Related Institutions (5)
University of Paris
174.1K papers, 5M citations

96% related

Centre national de la recherche scientifique
382.4K papers, 13.6M citations

93% related

ETH Zurich
122.4K papers, 5.1M citations

92% related

Imperial College London
209.1K papers, 9.3M citations

91% related

École Polytechnique Fédérale de Lausanne
98.2K papers, 4.3M citations

91% related

Performance
Metrics
No. of papers from the Institution in previous years
YearPapers
2023166
2022698
20215,126
20205,328
20195,192
20184,999