Angluin D.,Yale University |
Becerra-Bonache L.,CNRS Hubert Curien Laboratory
Artificial Intelligence | Year: 2017
We present a computational model that takes into account semantics for language learning and allows us to model meaning-preserving corrections. The model is constructed with a learner and a teacher who interact in a sequence of shared situations by producing utterances intended to denote a unique object in each situation. We test our model with limited sublanguages of 10 natural languages exhibiting a variety of linguistic phenomena. The results show that learning to a high level of performance occurs after a reasonable number of interactions. Comparing the effect of a teacher who does no correction to that of a teacher who corrects whenever possible, we show that under certain conditions corrections can accelerate the rate of learning. We also define and analyze a simplified model of a probabilistic process of collecting corrections to help understand the possibilities and limitations of corrections in our setting. © 2016 Elsevier B.V.
Combes C.,CNRS Hubert Curien Laboratory |
Azema J.,Jean Monnet University
Decision Support Systems | Year: 2013
The aim of this paper is to find feature-patterns related to the autonomy-disability level of elderly people living in nursing homes. These levels correspond to profiles based on the people's ability to perform activities of daily living like being able to wash, dress and move. To achieve this aim, an unsupervised approach is used. In this article, we propose a new clustering approach based on principal component analysis (PCA) to better approximate clusters. We want to automatically find categories or groups of residents based on their degree of autonomy-disability. All residents in a group have similar patterns. The main function of PCA is to explore the links between variables and the similarities between examples (individuals). The proposed algorithm uses the PCA technique to direct the determination of the clusters with self-organizing partitions by using the Euclidian distance. The study was carried out in close collaboration with the French cooperative health organization called the "Mutualité Française de la Loire". The quantitative data arises from the databases of four different nursing homes located in the city of Saint-Etienne in France. The study concerns 2271 observations of dependence evaluations corresponding to 628 residents. © 2012 Elsevier B.V.
Verrier N.,CNRS Hubert Curien Laboratory |
Fournier C.,CNRS Hubert Curien Laboratory
Optics Letters | Year: 2015
In-line digital holography (DH) is used in many fields to locate and size micro or nano-objects spread in a volume. To reconstruct simple shaped objects, the optimal approach is to fit an imaging model to accurately estimate their position and their characteristic parameters. Increasing the accuracy of the reconstruction is a big issue in DH, particularly when the pixel is large or the signal-to-noise ratio is low. We suggest exploiting the information redundancy of videos to improve the reconstruction of the holograms by jointly estimating the position of the objects and the characteristic parameters. Using synthetic and experimental data, we checked experimentally that this approach can improve the accuracy of the reconstruction by a factor more than the square root of the image number. © 2015 Optical Society of America.
Itina T.E.,CNRS Hubert Curien Laboratory |
Voloshko A.,CNRS Hubert Curien Laboratory
Applied Physics B: Lasers and Optics | Year: 2013
Recent promising methods of nanoparticle fabrication include laser ablation and spark discharge. Despite different experimental conditions, a striking similarity is often observed in the sizes of the obtained particles. To explain this result, we elucidate physical mechanisms involved in the formation of metallic nanoparticles. In particular, we compare supersaturation degree and sizes of critical nucleus obtained under laser ablation conditions with that obtained for spark discharge in air. For this, the dynamics of the expansion of either ablated or eroded products is described by using a three-dimensional blast wave model. Firstly, we consider nanosecond laser ablation in air. In the presence of a background gas, the plume expansion is limited by the gas pressure. Nanoparticles are mostly formed by nucleation and condensation taking place in the supersaturated vapor. Secondly, we investigate nanoparticles formation by spark discharge at atmospheric pressure. After efficient photoionization and streamer expansion, the cathode material suffers erosion and NPs appear. The calculation results allow us to examine the sizes of critical nuclei as function of the experimental parameters and to reveal the conditions favorable for the size reduction and for the increase in the nanoparticle yield. © 2013 Springer-Verlag Berlin Heidelberg.
Clackdoyle R.,CNRS Hubert Curien Laboratory
IEEE Transactions on Nuclear Science | Year: 2013
In medical imaging, a fanbeam projection refers to a collection of line integrals of some two-dimensional function, for lines converging on a point known as the source point for X-ray applications. In this work, necessary and sufficient consistency conditions (also known as range conditions) are given for fanbeam projections when the source trajectory follows an infinite straight line. The conditions are specified for both angular and linear parameterizations of the projection rays. In both cases, integrals of the fanbeam projections multiplied by a certain function will be a polynomial in the trajectory variable, so these consistency conditions can be considered true analogs of the well-known Helgason-Ludwig range conditions for parallel projections. © 1963-2012 IEEE.
Fernando B.,Catholic University of Leuven |
Fromont E.,CNRS Hubert Curien Laboratory |
Muselet D.,CNRS Hubert Curien Laboratory |
Sebban M.,CNRS Hubert Curien Laboratory
Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition | Year: 2012
Bag-of-words-based image classification approaches mostly rely on low level local shape features. However, it has been shown that combining multiple cues such as color, texture, or shape is a challenging and promising task which can improve the classification accuracy. Most of the state-of-the-art feature fusion methods usually aim to weight the cues without considering their statistical dependence in the application at hand. In this paper, we present a new logistic regression-based fusion method, called LRFF, which takes advantage of the different cues without being tied to any of them. We also design a new marginalized kernel by making use of the output of the regression model. We show that such kernels, surprisingly ignored so far by the computer vision community, are particularly well suited to achieve image classification tasks. We compare our approach with existing methods that combine color and shape on three datasets. The proposed learning-based feature fusion process clearly outperforms the state-of-the art fusion methods for image classification. © 2012 IEEE.
Bossuet L.,CNRS Hubert Curien Laboratory
Sustainable Computing: Informatics and Systems | Year: 2014
The electronics industry today is not yet green and/or sustainable. Indeed, the microelectronics industry is a consumer of primary materials, chemical products, water and energy. The manufacture of electronic products and their disposal at the end of their lives results in large quantities of waste products of varying degrees of toxicity that are difficult to deal with. Due to their high replacement rate, the lifespan of electronic products is spectacularly short. To reduce the environmental impact of electronic products the usual reduce-reuse-recycle (3R) trilogy appears to be insufficient. To achieve the objective of sustainable electronics, in this paper we suggest adding a fourth R for reconfigure. We recommend the use of the reconfiguration capacities of reconfigurable circuits such as FPGAs to reduce the functional obsolescence of electronic products by updating hardware. This paper is a survey of the sustainability of microelectronic. It presents some examples of pioneer works to illustrate the architecture of sustainable reconfigurable computing systems. © 2014 Elsevier Inc. All rights reserved.
Fernando B.,Catholic University of Leuven |
Fromont E.,CNRS Hubert Curien Laboratory |
Tuytelaars T.,Catholic University of Leuven
International Journal of Computer Vision | Year: 2014
Mid-level or semi-local features learnt using class-level information are potentially more distinctive than the traditional low-level local features constructed in a purely bottom-up fashion. At the same time they preserve some of the robustness properties with respect to occlusions and image clutter. In this paper we propose a new and effective scheme for extracting mid-level features for image classification, based on relevant pattern mining. In particular, we mine relevant patterns of local compositions of densely sampled low-level features. We refer to the new set of obtained patterns as Frequent Local Histograms or FLHs. During this process, we pay special attention to keeping all the local histogram information and to selecting the most relevant reduced set of FLH patterns for classification. The careful choice of the visual primitives and an extension to exploit both local and global spatial information allow us to build powerful bag-of-FLH-based image representations. We show that these bag-of-FLHs are more discriminative than traditional bag-of-words and yield state-of-the-art results on various image classification benchmarks, including Pascal VOC. © 2014 Springer Science+Business Media New York.
Fischer V.,CNRS Hubert Curien Laboratory
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2012
The issue of random number generation is crucial for the implementation of cryptographic systems. Random numbers are often used in key generation processes, authentication protocols, zeroknowledge protocols, padding, in many digital signature and encryption schemes, and even in some side channel attack countermeasures. For these applications, security depends to a great extent on the quality of the source of randomness and on the way this source is exploited. The quality of the generated numbers is checked by statistical tests. In addition to the good statistical properties of the obtained numbers, the output of the generator used in cryptography must be unpredictable. Besides quality and unpredictability requirements, the generator must be robust against aging effects and intentional or unintentional environmental variations, such as temperature, power supply, electromagnetic emanations, etc. In this paper, we discuss practical aspects of a true random number generator design. Special attention is given to the analysis of security requirements and on the way how this requirements can be met in practice. © 2012 Springer-Verlag.
Itina T.E.,CNRS Hubert Curien Laboratory
Journal of Physical Chemistry C | Year: 2011
Nanoparticle formation by laser ablation in liquids is studied numerically. We investigate such processes as cluster ejection, cluster nucleation, and aggregation. First, laser plume formation, its expansion, and confinement by the liquid are considered. These processes are connected with the formation of two shock waves: one moving inside the solid and the second one propagating in the liquid. If short and ultrashort laser pulses are used, the created plasma plume does not absorb laser radiation. In this case, a larger energy fraction is transferred into the solid during much shorter time, so that the ablation process is explosive. As a result, cluster precursors are ejected directly from the target, and the created plasma is confined to a smaller volume. Shorter laser pulses thus provide more favorable conditions for nanoparticle formation. In addition to the following short plasma expansion stage, a much longer aggregation process occurring in the liquid phase is found to affect the final size distribution. © 2010 American Chemical Society.