The two steps of the proposed automatic path planner are controlled thanks to the high abstraction level information involved in our **multi**- **layer** environment model. This control aims at improving the path planning process performances from both a qualitative (relevance of paths) and quantitative (succes rate and computation time) point of view. The coarse planning step deﬁnes a topological path within the environment. The semantic information is used to chose the topological path with higher relevance. When deﬁning the actual geometrical tra- jectory during the second step of ﬁne planning, diﬀerent path planning algorithms can be used depending on the local semantic information associated with the place belonging to the topological step considered. 5. Implementation

En savoir plus
2 Normandie Univ., ENSICAEN, CNRS, GREYC
julien.rabin@unicaen.fr, sites.google.com/site/rabinjulien
Abstract. The optimal transport (OT) framework has been largely used in inverse imaging and computer vision problems, as an interest- ing way to incorporate statistical constraints or priors. In recent years, OT has also been used in machine learning, mostly as a metric to com- pare probability distributions. This work addresses the semi-discrete OT problem where a continuous source distribution is matched to a discrete target distribution. We introduce a fast stochastic algorithm to approx- imate such a semi-discrete OT problem using a hierarchical **multi**-**layer** transport plan. This method allows for tractable computation in high- dimensional case and for large point-clouds, both during training and synthesis time. Experiments demonstrate its numerical advantage over **multi**-scale (or **multi**-level) methods. Applications to fast exemplar-based texture synthesis based on patch matching with two layers, also show stunning improvements over previous single **layer** approaches. This shal- low model achieves comparable results with state-of-the-art deep learning methods, while being very compact, faster to train, and using a single image during training instead of a large dataset.

En savoir plus
Figure 17. Convergence history of objective function and the volume constraints.
5. Conclusions and Discussion
The problem of an optimal **multi**-**layer** micro-structure is considered. We use inverse ho- mogenization, the Hadamard shape derivative and a level set method to track boundary changes, within the context of the smooth interface, in the periodic unit cell. We produce several examples of auxetic micro-structures with different volume constraints as well as different ways of enforcing the aforementioned constraints. The **multi**-**layer** interpretation suggests a particular way on how to approach the subject of 3D printing the micro-structures. The magenta material is essentially the cyan material layered twice producing a small ex- trusion with the process repeated several times. This **multi**-**layer** approach has the added benefit that some of the contact among the material parts is eliminated, thus allowing the structure to be further compressed than if the material was in the same plane.

En savoir plus
Abstract—Wireless sensor networks (WSNs) have become
pervasive and are used for a plethora of applications and services. They are usually deployed with specific applications and services; thereby precluding their re-use when other applications and services are contemplated. This can inevitably lead to the proliferation of redundant WSN deployments. Virtualization is a technology that can aid in tackling this issue. It enables the sharing of resources/infrastructures by multiple independent entities. This position paper proposes a novel **multi**-**layer** architecture for WSN virtualization and identifies the research challenges. Related work is also discussed. We illustrate the potential of the architecture by applying it to a scenario in which WSNs are shared for fire monitoring.

En savoir plus
High energy PIXE is a useful and non-detructive tool to characterize **multi**- **layer** thick samples such as cultural heritage objects. In a previous work, we demonstrated the possibility to perform quantitative analysis of simple **multi**-**layer** samples using high energy PIXE, without any assumption on their composition. In this work an in-depth study of the parameters involved in the method previously published is proposed. Its extension to more complex samples with a repeated **layer** is also presented. Experiments have been per- formed at the ARRONAX cyclotron using 68 MeV protons. The thicknesses and sequences of a **multi**-**layer** sample including two different layers of the same element have been determined. Performances and limits of this method are presented and discussed.

En savoir plus
I. I NTRODUCTION
S PARSITY has been at the heart of a plethora of signal
processing and data analysis techniques over the last two decades. These techniques usually impose that the objects of interest be sparse in a certain domain. They owe their success to the fact that sparse objects are easier to manipulate and more prone to interpretation than dense ones especially in high dimension. However, to efficiently manipulate high- dimensional data, it is not sufficient to rely on sparse objects: efficient operators are also needed to manipulate these objects. The n-dimensional Discrete Fourier Transform (DFT) is certainly the most well known linear operator with an efficient implementation: the Fast Fourier Transform (FFT) [3], allows to apply the operator in O(n log n) arithmetic operations instead of O(n 2 ) in its dense form. Similar complexity savings have been achieved for other widely used operators such as the Hadamard transform [4], the Discrete Cosine Transform (DCT) [5] or the Discrete Wavelet Transform (DWT) [6]. For all these fast linear transforms, the matrix A corresponding to the dense form of the operator admits a **multi**-**layer** sparse expression,

En savoir plus
background, thus enabling us to use distinct segmen- tation models for the input images. In the remaining of this paper, we will refer to this problem by “**Multi**- **Layer** Joint Segmentation” (MLJS). Table 1 summa- rizes the identified existing strategies for jointly seg- menting coregistered images. In what follows, we pro- vide some insights about each strategy. In [21], the au- thors introduce a **multi**-**layer** and **multi**-label Markov Random Field (MRF) model combining texture and color features for segmenting a single color image. The MRF involves three layers: one **layer** per feature (cor- responding to individual segmentations) and one **layer** for their indirect combination (corresponding to joint segmentation). This model includes standard spatial re- lationships between pixels per individual segmentation as well as relationships between pixels of individual seg- mentations and the joint segmentation. The intra-**layer** relationships ensure the regularity within each individ- ual segmentation whereas the inter-**layer** ones penal- ize any deviation between each individual segmentation and the joint one. Nevertheless, the energy term enforc- ing the consistency between individual segmentations is not well defined for some pixel pairs. Furthermore, in the original work, this MRF is suboptimally solved us- ing Iterated Conditional Modes (ICM), while the same MRF is optimally solved using graph cuts (GC) in [4] but only for two labels and for a single image pair.

En savoir plus
segmentation and annotation
Thomas Dietenbeck, Fakhri Torkhani, Ahlem Othmani, Marco Attene, Jean-Marie Favreau
Abstract Mesh segmentation and semantic annotation are used as preprocessing steps for many applications, including shape retrieval, mesh abstraction, and adap- tive simplification. In current practice, these two steps are done sequentially: a purely geometrical analysis is employed to extract the relevant parts, and then these parts are annotated. We introduce an original framework where annotation and seg- mentation are performed simultaneously, so that each of the two steps can take ad- vantage of the other. Inspired by existing methods used in image processing, we employ an expert’s knowledge of the context to drive the process while minimiz- ing the use of geometric analysis. For each specific context a **multi**-**layer** ontology can be designed on top of a basic knowledge **layer** which conceptualizes 3D object features from the point of view of their geometry, topology, and possible attributes. Each feature is associated with an elementary algorithm for its detection. An expert can define the upper layers of the ontology to conceptualize a specific domain with- out the need to reconsider the elementary algorithms. This approach has a twofold advantage: on one hand it allows to leverage domain knowledge from experts even if they have limited or no skills in geometry processing and computer program-

En savoir plus
Calculation of heat conduction transfer functions for **multi**-**layer** slabs
Stephenson, D. G.; Mitalas, G. P.
https://publications-cnrc.canada.ca/fra/droits
L’accès à ce site Web et l’utilisation de son contenu sont assujettis aux conditions présentées dans le site LISEZ CES CONDITIONS ATTENTIVEMENT AVANT D’UTILISER CE SITE WEB.

diagnostic, drug delivery, tissue regeneration and photothermal cancer ablation. In view of the enormous development of graphene-based technologies, a careful assessment of its impact on health and environment is demanded. It is evident how investigating the graphene toxicity is of fundamental importance in the context of medical purposes. On the other hand, the nanomaterial present in the environment, likely to be generated all along the industrial life-cycle, may have harmful effects on living organisms. In the present work, an important contribution on the impact of **multi**-**layer** graphene (MLG) on health and environment is given by using a multifaceted approach. For the ﬁrst purpose, the effect of the material on two mammalian cell models was assessed. Key cytotoxicity parameters were considered such as cell viability and inﬂammatory response induction. This was combined with an evaluation of MLG toxicity towards Xenopus laevis, used as both in vivo and environmental model organism.

En savoir plus
A **Multi**-**layer** Framework for Virtual Organizations Creation 315
4. Related Work
There are many ongoing research efforts related to the VO creation framework within a VO Breeding Environment context. The most relevant work in this area is proposed by Camarinah-Matos et al. within the ECOLEAD project [1]. This work characterizes a first attempt to the definition of a reference framework for VO creation that addresses the fundamental elements of the VBEs. The proposed framework presents a detailed process covering a number of steps from the identification of the collaboration opportunity till the lunching of the VO. Besides, the framework is validated through empirical studies relating to international industry based VBE involved in the ECOLEAD project.

En savoir plus
In many natural and engineered systems, the interactions be- tween sets of variables in different subsystems involve multiple layers of interdependencies. This is for instance the case in the neural networks developed in deep learning [1], the hierarchi- cal models used in statistical inference [2], and the multiplex networks considered in complex systems [3]. It is therefore fundamental to generalize our theoretical and algorithmic tools to deal with these **multi**-**layer** setups. Our goal in this paper is to develop such a generalization of the cavity/replica approach that originated in statistical physics [4] and that has been shown to be quite successful for studying generalized linear estimation with randomly chosen mixing, leading in particular to the computation of the mutual information (or equivalently free energy) and minimum achievable mean-squared error for CDMA and compressed sensing [5]–[8]. This methodology is also closely related to the approximate message passing (AMP) algorithm, originally known in physics as Thouless- Anderson-Palmer (TAP) equations [9]–[14].

En savoir plus
Advances in Wireless and Mobile Communication Networks: A **Multi**-**Layer** Analysis
Prénom, Nom : Sihame EL-HAMMANI
Abstract:
The aim of this thesis is to provide the reader with a comprehensive view of the advances in wireless and mobile communication networks, while adopting a **multi**-**layer** analysis. We rely on a TCP/IP architecture, and by browsing layers we try to cover different aspects of analysis, design, and optimization problems for heterogeneous wireless networks. We are more particularly interested in the physic, link and application layers that are very important, play a major role and have a great influence in the system performances. In the physical **layer**, we study and analyse the performance of two new technologies. Firstly, the wireless metropolitan area network (MAN) illustrated in WiMAX technology. And secondly, femtocell technology as special class of small cells, that can cover small areas with low power transmission which has become increasingly important for the indoor areas. Furthermore, as the choice of medium access control (MAC) protocol can affect the system performance and the use of wireless networks, we investigate one of the most popular random accesses MAC protocols, illustrated on slotted aloha protocol that affects directly the interoperability and the global roaming of mobile users. And finally, we focus on a practical case in the application **layer** by addressing the minimization of content diffusion time problem in social networks.

En savoir plus
172 En savoir plus

Email: jml@math.tamu.edu
Abstract—We study a deep/**multi**-**layer** structured matrix fac- torization problem. It approximates a given matrix by the product of K matrices (called factors). Each factor is obtained by applying a fixed linear operator to a short vector of parameters (thus the name ”structured”). We call the model deep or **multi**- **layer** because the number of factors is not limited. In the practical situations we have in mind, we typically have K = 10 or 20.

In order to improve the transmission reliability in current wireless communication systems, the Hybrid Automatic ReQuest (HARQ) protocol is employed to manage the unknown time-varying channel. The acknowledgments are fed back with delay on the return link. To fill up the idle time between a transmission and its acknowledgment, parallel HARQ streams associated with different messages are carried out. In this paper we improve on parallel HARQ by proposing a **multi**-**layer** HARQ protocol (also called superposition coding or **multi**-packet HARQ), where a single transmission may carry information on multiple messages. The **multi**-**layer** HARQ protocol works in presence of delay on the return link as parallel HARQ does, and does not require additional feedback such as the channel state information. It aims at improving the accuracy as well as the user’s delay distribution, thus achieving throughput increase. Assuming capacity-achieving codes, we show that the proposed protocol outperforms parallel HARQ in throughput, message error rate, and delay distribution. Using practical codes and decoding algorithms the gains are as well significant, at the expense of the receiver’s complexity.

En savoir plus
Y = W C T
Fig. 1. Scheme of a feed-forward **multi**-**layer** perceptron.
For connectionist modeling, the architecture of the network as well as the algorithms of minimization andoptimization needto be speci5ed. Of the many connectionist mod- els, such as Radial Basis Function networks or wavelet networks, the **multi**-**layer** per- ceptron (MLP) is one that has been extremely useful in many applications andhas been extensively analyzedtheoretically. In this article, we focusedon the MLP model that is the most commonly used connectionist model in the medical 5eld. But a further step in this work will be to compare other neural network models to classical statisti- cal models in order to obtain a more exhaustive approach. We performed a perceptron model with one hidden **layer**, and a learning algorithm based on the delta-rule (see

En savoir plus
MIT Media Lab, Building E14, 75 Amherst St. Cambridge, MA, 02139, USA E-mail: mhirsch@media.mit.edu
Abstract. Near-term commercial **multi**-view displays currently employ ray-based 3D or 4D light field techniques. Conventional approaches to ray-based display typically include lens arrays or heuristic barrier patterns combined with integral interlaced views on a display screen such as an LCD panel. Recent work has placed an emphasis on the co-design of optics and image formation algorithms to achieve increased frame rates, brighter images, and wider fields-of-view using optimization-in-the-loop and novel arrangements of commodity LCD panels. In this paper we examine the construction and calibration methods of computational, **multi**-**layer** LCD light field displays. We present several experimental configurations that are simple to build and can be tuned to sufficient precision to achieve a research quality light field display. We also present an analysis of moir´ e interference in these displays, and guidelines for diffuser placement and display alignment to reduce the effects of moir´ e. We describe a technique using the moir´ e magnifier to fine-tune the alignment of the LCD layers.

En savoir plus
between different structures, wideband relative phase shifts could be obtained with a wide phase dynamic range. The main disadvantage of this structure was that it had relatively large insertion loss values, especially in applications requiring many phase shifters as beam-forming matrices. These insertion loss values were however within the typical range of non-lumped elements based CRLH implementations. Then, to exploit the **multi**-**layer** SIW topology, a novel wideband, low-loss two-**layer** transition was developed. The transition was based on a transverse narrow slot coupling two short-ended waveguides in two different layers. According to a wideband equivalent circuit model, a detailed study of the transmission phase and magnitude of the transition was carried out. Based on its equivalent circuit model, the transition could be optimized within the well known unequal width, equal-length SIW phase shifters in order to compensate its additional phase shift within the frequency band of interest. This two-**layer** wideband phase shifter scheme was adopted in the final developed matrix architecture. On the other hand, the low-loss wideband interconnect was then exploited to develop a three-**layer**, multiply-folded waveguide structure as a good candidate for compensated-length, variable width, low-loss, compact wideband phase shifters in SIW technology.

En savoir plus
128 En savoir plus

45 ° , and between δ = 0.037 and 0.052 for θ o = 30 ° .
Conclusion
The steady state laminar natural convection within **multi**-**layer** domes heated from the outside was investigated for different dome shapes. The flow pattern and temperature field within the gap between layers and the convection heat transfer were presented. Correlations were developed for the convection heat transfer as a function of the dome shape and the gap spacing.

Context{dbU serN ame : anonyme; IpAddress : 192.000.111.0}
With the context information available, the process to find the dependencies be- tween policies is described in Algorithm 1. Basically, for each context parameter in a given policy it searches if there is any rule using that attribute value in any of the other policies. If this is the case, a dependency exists between both policies and as such it is registered. Note that the algorithm has been optimized by considering that no circular dependencies exist. The set of candidate policies for a Policy P j (i.e., policies it may depend on), initialized to all the other policies in line 3, is modified in line 13 to re- move Policies P i that already depend on it. This assumption stems from the nature of **multi**-**layer** ISs where upper components depend only on components in lower layers. This optimization can be dropped for other scenarios if needed.

En savoir plus