scispace - formally typeset
Search or ask a question
Institution

École Polytechnique Fédérale de Lausanne

FacilityLausanne, Switzerland
About: École Polytechnique Fédérale de Lausanne is a facility organization based out in Lausanne, Switzerland. It is known for research contribution in the topics: Population & Catalysis. The organization has 44041 authors who have published 98296 publications receiving 4372092 citations. The organization is also known as: EPFL & ETHL.


Papers
More filters
Book ChapterDOI
08 Oct 2016
TL;DR: This work proposes to augment feedforward nets for object segmentation with a novel top-down refinement approach that is capable of efficiently generating high-fidelity object masks and is 50 % faster than the original DeepMask network.
Abstract: Object segmentation requires both object-level information and low-level pixel data. This presents a challenge for feedforward networks: lower layers in convolutional nets capture rich spatial information, while upper layers encode object-level knowledge but are invariant to factors such as pose and appearance. In this work we propose to augment feedforward nets for object segmentation with a novel top-down refinement approach. The resulting bottom-up/top-down architecture is capable of efficiently generating high-fidelity object masks. Similarly to skip connections, our approach leverages features at all layers of the net. Unlike skip connections, our approach does not attempt to output independent predictions at each layer. Instead, we first output a coarse ‘mask encoding’ in a feedforward pass, then refine this mask encoding in a top-down pass utilizing features at successively lower layers. The approach is simple, fast, and effective. Building on the recent DeepMask network for generating object proposals, we show accuracy improvements of 10–20% in average recall for various setups. Additionally, by optimizing the overall network architecture, our approach, which we call SharpMask, is 50 % faster than the original DeepMask network (under .8 s per image).

823 citations

Journal ArticleDOI
TL;DR: These updated recommendations take into account all rTMS publications, including data prior to 2014, as well as currently reviewed literature until the end of 2018, and are based on the differences reached in therapeutic efficacy of real vs. sham rT MS protocols.

822 citations

Journal ArticleDOI
TL;DR: In this paper, the authors studied the promotional effects of certain transition metal ions on the activity of amorphous MoS3 films and found that Fe, Co, and Ni ions promote the growth of the MoS-3 films, resulting a high surface area and a higher catalyst loading.
Abstract: Molybdenum sulfide materials have been shown as promising non-precious catalysts for hydrogen evolution. This paper describes the study of the promotional effects of certain transition metal ions on the activity of amorphous MoS3 films. Ternary metal sulfide films, M–MoS3 (M = Mn, Fe, Co, Ni, Cu, Zn), have been prepared by cyclic voltammetry of aqueous solutions containing MCl2 and (NH4)2[MoS4]. Whereas the Mn–, Cu–, and Zn–MoS3 films show similar or only slightly higher catalytic activity as the MoS3 film, the Fe–, Co–, and Ni–MoS3 films are significantly more active. The promotional effects of Fe, Co, and Ni ions exist under both acidic and neutral conditions, but the effects are more pronounced under neutral conditions. Up to a 12-fold increase in exchange current density and a 10-fold increase in the current density at an overpotential of 150 mV are observed at pH = 7. It is shown that Fe, Co, and Ni ions promote the growth of the MoS3 films, resulting a high surface area and a higher catalyst loading. These changes are the main contributors to the enhanced activity at pH = 0. However, at pH = 7, Fe, Co, and Ni ions appear to also increase the intrinsic activity of the MoS3 film.

821 citations

Journal ArticleDOI
TL;DR: This paper gives an overview of the most prominent methods for evolving ANNs with a special focus on recent advances in the synthesis of learning architectures.
Abstract: Artificial neural networks (ANNs) are applied to many real-world problems, ranging from pattern clas- sification to robot control. In order to design a neural network for a particular task, the choice of an architecture (including the choice of a neuron model), and the choice of a learning algorithm have to be addressed. Evolutionary search methods can provide an automatic solution to these problems. New insights in both neuroscience and evolu- tionary biology have led to the development of increasingly powerful neuroevolution techniques over the last decade. This paper gives an overview of the most prominent methods for evolving ANNs with a special focus on recent advances in the synthesis of learning architectures.

821 citations

Journal ArticleDOI
TL;DR: Angular and spectrally resolved luminescence show that the polariton emission is beamed in the normal direction with an angular width of +/-5 degrees and spatial size around 5 microm.
Abstract: We observe a room-temperature low-threshold transition to a coherent polariton state in bulk GaN microcavities in the strong-coupling regime. Nonresonant pulsed optical pumping produces rapid thermalization and yields a clear emission threshold of 1 mW, corresponding to an absorbed energy density of 29 mu J cm(-2), 1 order of magnitude smaller than the best optically pumped (In,Ga)N quantum-well surface-emitting lasers (VCSELs). Angular and spectrally resolved luminescence show that the polariton emission is beamed in the normal direction with an angular width of +/- 5 degrees and spatial size around 5 mu m.

820 citations


Authors

Showing all 44420 results

NameH-indexPapersCitations
Michael Grätzel2481423303599
Ruedi Aebersold182879141881
Eliezer Masliah170982127818
Richard H. Friend1691182140032
G. A. Cowan1592353172594
Ian A. Wilson15897198221
Johan Auwerx15865395779
Menachem Elimelech15754795285
A. Artamonov1501858119791
Melody A. Swartz1481304103753
Henry J. Snaith146511123155
Kurt Wüthrich143739103253
Richard S. J. Frackowiak142309100726
Jean-Paul Kneib13880589287
Kevin J. Tracey13856182791
Network Information
Related Institutions (5)
ETH Zurich
122.4K papers, 5.1M citations

98% related

Massachusetts Institute of Technology
268K papers, 18.2M citations

96% related

Georgia Institute of Technology
119K papers, 4.6M citations

96% related

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

96% related

University of Illinois at Urbana–Champaign
225.1K papers, 10.1M citations

94% related

Performance
Metrics
No. of papers from the Institution in previous years
YearPapers
2023234
2022704
20215,247
20205,644
20195,432
20185,094