Search results for "DEEP"
showing 10 items of 1434 documents
Neutron scattering and imaging: a tool for archaeological studies
2015
International audience; Neutron scattering and neutron imaging are powerful techniques for studying the structure of objects without damage, which is an essential prerequisite for investigations in Cultural Heritage domain, particularly in Archaeology. The deep penetration of neutrons in most materials allows for the study of relatively large objects. The contrast between similar materials, like metals in alloys, or that due to the presence of hydrogen atoms gives information about the internal structure of objects that have been modified or repaired in the past. Imaging and tomography give a 3-dimensional view of the whole object, permitting discrimination between different parts of the ob…
Towards Open Domain Chatbots—A GRU Architecture for Data Driven Conversations
2018
Understanding of textual content, such as topic and intent recognition, is a critical part of chatbots, allowing the chatbot to provide relevant responses. Although successful in several narrow domains, the potential diversity of content in broader and more open domains renders traditional pattern recognition techniques inaccurate. In this paper, we propose a novel deep learning architecture for content recognition that consists of multiple levels of gated recurrent units (GRUs). The architecture is designed to capture complex sentence structure at multiple levels of abstraction, seeking content recognition for very wide domains, through a distributed scalable representation of content. To …
Statistical retrieval of atmospheric profiles with deep convolutional neural networks
2019
Abstract Infrared atmospheric sounders, such as IASI, provide an unprecedented source of information for atmosphere monitoring and weather forecasting. Sensors provide rich spectral information that allows retrieval of temperature and moisture profiles. From a statistical point of view, the challenge is immense: on the one hand, “underdetermination” is common place as regression needs to work on high dimensional input and output spaces; on the other hand, redundancy is present in all dimensions (spatial, spectral and temporal). On top of this, several noise sources are encountered in the data. In this paper, we present for the first time the use of convolutional neural networks for the retr…
Tree Species Classification of Drone Hyperspectral and RGB Imagery with Deep Learning Convolutional Neural Networks
2020
Interest in drone solutions in forestry applications is growing. Using drones, datasets can be captured flexibly and at high spatial and temporal resolutions when needed. In forestry applications, fundamental tasks include the detection of individual trees, tree species classification, biomass estimation, etc. Deep neural networks (DNN) have shown superior results when comparing with conventional machine learning methods such as multi-layer perceptron (MLP) in cases of huge input data. The objective of this research is to investigate 3D convolutional neural networks (3D-CNN) to classify three major tree species in a boreal forest: pine, spruce, and birch. The proposed 3D-CNN models were emp…
Understanding deep learning in land use classification based on Sentinel-2 time series
2020
AbstractThe use of deep learning (DL) approaches for the analysis of remote sensing (RS) data is rapidly increasing. DL techniques have provided excellent results in applications ranging from parameter estimation to image classification and anomaly detection. Although the vast majority of studies report precision indicators, there is a lack of studies dealing with the interpretability of the predictions. This shortcoming hampers a wider adoption of DL approaches by a wider users community, as model’s decisions are not accountable. In applications that involve the management of public budgets or policy compliance, a better interpretability of predictions is strictly required. This work aims …
Zr/Hf ratio and REE behaviour: A coupled indication of lithogenic input in marginal basins and deep-sea brines
2019
Abstract The distribution of dissolved Zr, Hf and Rare Earth Elements (yttrium and lanthanides, hereafter referred to as REE) in the Eastern Mediterranean seawater column was measured in the Kryos basin to evaluate the lithogenic contribution from both Nile River and Sahara and Arabian desert dust. We found dissolved Zr/Hf ratios below the signature of crustal rocks and chondrites; a phenomenon likely driven by the dissolution of the Mn-rich coating of atmospheric dust particles delivered from the desert. In deeper waters, Zr/Hf ratios are clustered close to the signature of crustal rocks and chondrites according to the different Zr and Hf dissolved speciation. The Zr/Hf ratio observed in t…
Transferring deep learning models for cloud detection between Landsat-8 and Proba-V
2020
Abstract Accurate cloud detection algorithms are mandatory to analyze the large streams of data coming from the different optical Earth observation satellites. Deep learning (DL) based cloud detection schemes provide very accurate cloud detection models. However, training these models for a given sensor requires large datasets of manually labeled samples, which are very costly or even impossible to create when the satellite has not been launched yet. In this work, we present an approach that exploits manually labeled datasets from one satellite to train deep learning models for cloud detection that can be applied (or transferred) to other satellites. We take into account the physical proper…
Active Degassing of Deeply Sourced Fluids in Central Europe: New Evidences From a Geochemical Study in Serbia
2021
We report on the results of an extensive geochemical survey of fluids released in the Vardar zone (central-western Serbia), a mega-suture zone at the boundary between Eurasia and Africa plates. Thirty-one bubbling gas samples are investigated for their chemical and isotopic compositions (He, C, Ar) and cluster into three distinct groups (CO2-dominated, N2-dominated, and CH4-dominated) based on the dominant gas species. The measured He isotope ratios range from 0.08 to 1.19 Ra (where Ra is the atmospheric ratio), and reveal for the first time the presence of a minor (<20%) but detectable regional mantle-derived component in Serbia. δ13C values range from −20.2‰ to −0.1‰ (versus PDB), with…
Numerical evidence for thermohaline circulation reversals during the Maastrichtian
2005
[1] The sensitivity of the Maastrichtian thermohaline circulation to the opening/closing of marine communications between the Arctic and North Pacific oceans is investigated through a set of numerical experiments using the model CLIMBER-2 (Earth Model of Intermediate Complexity). We show here that the opening or closing of an Arctic-Pacific marine gateway induces transitions between different equilibrium states of the thermohaline circulation. Sensitivity tests of the inferred modes of thermohaline circulation to atmospheric CO2 level changes have also been explored. An abrupt switch in deep convection from high northern to high southern latitudes, a change consistent with isotopic evidence…
Radiogenic isotopes: new tools help reconstruct paleocean circulation and erosion input
2001
Ocean and atmosphere circulation and continental weathering regimes have undergone great changes over thousands of years as well as tens of millions of years. During the glacial stages of the Pleistocene, ocean circulation was generally more sluggish and deep water circulation in the Atlantic had a shallower flow. At the same time, weathering on the continents was enhanced by glacial erosion, particularly in high northern latitudes, which increased the input of erosional detritus into the ocean. In addition, atmospheric pressure gradients were larger, leading to higher wind speeds and increased supply of aeolian dust to the ocean. Prior to the onset of Northern Hemisphere glaciation and pro…