• Publicationsclose
• Other research productsclose
• Preprint close
• FR close
• Hal-Diderot close
• Digital Humanities and Cultural Heritage close

In some contexts, well-formed natural language cannot be expected as input to information or communication systems. In these contexts, the use of grammar-independent input (sequences of uninflected semantic units like e.g. language-independent icons) can be an answer to the users' needs. A semantic analysis can be performed, based on lexical semantic knowledge: it is equivalent to a dependency analysis with no syntactic or morphological clues. However, this requires that an intelligent system should be able to interpret this input with reasonable accuracy and in reasonable time. Here we propose a method allowing a purely semantic-based analysis of sequences of semantic units. It uses an algorithm inspired by the idea of ``chart parsing'' known in Natural Language Processing, which stores intermediate parsing results in order to bring the calculation time down. In comparison with using declarative logic programming - where the calculation time, left to a prolog engine, is hyperexponential -, this method brings the calculation time down to a polynomial time, where the order depends on the valency of the predicates. Comment: 7 pages, 1 figure, LaTeX 2e using COLACL and EPSF packages. Proceedings of the 19th International Conference on Computational Linguistics (COLING 2002), Taipei, Republic of China (Taiwan), 24 Aug. - 1 Sept. 2002

International audience; The main contribution of this paper is a new method for classifying document images by combining textual features extracted with the Bag of Words (BoW) technique and visual features extracted with the Bag of Visual Words (BoVW) technique. The BoVW is widely used within the computer vision community for scene classification or object recognition but few applications for the classification of entire document images have been submitted. While previous attempts have been showing disappointing results by combining visual and textual features with the Borda-count technique, we're proposing here a combination through learning approach. Experiments conducted on a 1925 document image industrial database reveal that this fusion scheme significantly improves the classification performances. Our concluding contribution deals with the choosing and tuning of the BoW and/or BoVW techniques in an industrial context.

Reconstruction of images from noisy linear measurements is a core problem in image processing, for which convex optimization methods based on total variation (TV) minimization have been the long-standing state-of-the-art. We present an alternative probabilistic reconstruction procedure based on approximate message-passing, Scampi, which operates in the compressive regime, where the inverse imaging problem is underdetermined. While the proposed method is related to the recently proposed GrAMPA algorithm of Borgerding, Schniter, and Rangan, we further develop the probabilistic approach to compressive imaging by introducing an expectation-maximizaiton learning of model parameters, making the Scampi robust to model uncertainties. Additionally, our numerical experiments indicate that Scampi can provide reconstruction performance superior to both GrAMPA as well as convex approaches to TV reconstruction. Finally, through exhaustive best-case experiments, we show that in many cases the maximal performance of both Scampi and convex TV can be quite close, even though the approaches are a prori distinct. The theoretical reasons for this correspondence remain an open question. Nevertheless, the proposed algorithm remains more practical, as it requires far less parameter tuning to perform optimally. Comment: Presented at the 2015 International Meeting on High-Dimensional Data Driven Science, Kyoto, Japan

Out-of-vocabulary (OOV) words can pose serious challenges for machine translation (MT) tasks, and in particular, for low-resource language (LRL) pairs, i.e., language pairs for which few or no parallel corpora exist. Our work adapts variants of seq2seq models to perform transduction of such words from Hindi to Bhojpuri (an LRL instance), learning from a set of cognate pairs built from a bilingual dictionary of Hindi--Bhojpuri words. We demonstrate that our models can be effectively used for language pairs that have limited parallel corpora; our models work at the character level to grasp phonetic and orthographic similarities across multiple types of word adaptations, whether synchronic or diachronic, loan words or cognates. We describe the training aspects of several character level NMT systems that we adapted to this task and characterize their typical errors. Our method improves BLEU score by 6.3 on the Hindi-to-Bhojpuri translation task. Further, we show that such transductions can generalize well to other languages by applying it successfully to Hindi -- Bangla cognate pairs. Our work can be seen as an important step in the process of: (i) resolving the OOV words problem arising in MT tasks, (ii) creating effective parallel corpora for resource-constrained languages, and (iii) leveraging the enhanced semantic knowledge captured by word-level embeddings to perform character-level tasks. Comment: 47 pages, 4 figures, 21 tables (including Appendices)

Whenever a compound exhibits a spontaneous muSR oscillation, long-range magnetic ordering is usually inferred. Here we show that some caution is required. The coherence length needs not to be large for a spontaneous muon spin precession to be observed. Establishing the incommensurate nature of a magnetic structure, solely based on muSR measurements, may not be reliable. The absence of a spontaneous muon precession at low temperature does not mean that the system under investigation does not display long-range magnetic ordering. The relaxation measured in zero and longitudinal field in the quasi-static limit is usually analyzed in the framework of the strong-collision model, the static polarization function being taken to be the famous Kubo-Toyabe function. This might not be satisfactory if short-range correlation effects are strong. Here we propose a method based on the maximum entropy concept and reverse Monte Carlo technique which gives results consistent with those obtained in 2013 by analytical means for the considered example. Comment: To appear in the proceedings of the 13th International Conference on Muon Spin Rotation, Relaxation and Resonance, Grindelwald, Switzerland, 1-6 June 2014

International audience; We present a bi-confluent Heun potential for the Schrödinger equation involving inverse fractional powers and a repulsive centrifugal-barrier term the strength of which is fixed to a constant. This is an infinite potential well defined on the positive half-axis. Each of the fundamental solutions for this conditionally integrable potential is written as an irreducible linear combination of two Hermite functions of a shifted and scaled argument. We present the general solution of the problem, derive the exact energy spectrum equation and construct a highly accurate approximation for the bound-state energy levels.

This paper presents the systems developed by LIUM and CVC for the WMT16 Multimodal Machine Translation challenge. We explored various comparative methods, namely phrase-based systems and attentional recurrent neural networks models trained using monomodal or multimodal data. We also performed a human evaluation in order to estimate the usefulness of multimodal data for human machine translation and image description generation. Our systems obtained the best results for both tasks according to the automatic evaluation metrics BLEU and METEOR. Comment: 7 pages, 2 figures, v4: Small clarification in section 4 title and content

We present MadAnalysis 5, an analysis package dedicated to phenomenological studies of simulated collisions occurring in high-energy physics experiments. Within this framework, users are invited, through a user-friendly Python interpreter, to implement physics analyses in a very simple manner. A C++ code is then automatically generated, compiled and executed. Very recently, the expert mode of the program has been extended so that analyses with multiple signal/control regions can be handled. Additional observables have also been included, and an interface to several fast detector simulation packages has been developed, one of them being a tune of the Delphes 3 software. As a result, a recasting of existing ATLAS and CMS analyses can be achieved straightforwardly. Comment: 5 pages, 3 figures, 1 table; submitted to the proceedings of the 16th International workshop on Advanced Computing and Analysis Techniques in physics research (ACAT 2014), Prague

c’est au XVIIIe siècle que naissent les notions grammaticales de «circonstanciel » sous la plume de l’abbé gabriel girard en 1747, et de «circonstances » (obstojatel′ stva) dans la Rossijskaja grammatika de Mixail Lomonosov en 1755. Pour Lomonosov, le mode d’expression privilégié des circonstances est le syntagme prépositionnel, et l’adverbe en est l’expression abrégée. Il ne parle pas des adverbes ou des groupes prépositionnels comme d’éléments modifiant le verbe, mais il se place au niveau de l’énoncé, présentant les circonstances comme l’expansion de la phrase ; il faudra attendre plus de deux siècles pour que cette caractéristique des circonstants soit prise en compte par les linguistes. Pour ce qui est de l’ordre des mots, chez l’abbé girard c’est le «feu de l’imagination » , pour Lomonosov c’est le «libre cours des pensées humaines » qui permettent de modifier en russe l’ordre des mots canonique / sujet – verbe/. L’analyse de la langue de n. karamzin montre que ce n’est ni le «feu de [ son] imagination » , ni le «libre cours de [ ses] pensées » qui gèrent le choix de l’ordre des mots chez lui, mais plutôt le désir et la nécessité qu’il éprouve de produire un texte cohérent, dans lequel les idées s’enchaînent de la manière la plus satisfaisante et la moins coûteuse pour l’énonciataire et cela lui permet de placer le circonstant en tête de phrase s’il a besoin de poser le cadre spatio-temporel de son énoncé, que les «circonstances » soient connues ou non de l’énonciataire. The Grammatical notion of “ circumstantial” appears in the 18th century in the writings of Abbé Gabriel Girard in 1747 and “ circumstances” (obstojatel′ stva) in Mixail Lomonosov’s Rossijskaja grammatika in 1755. For Lomonosov, the prepositional phrase is the best mode of expression for the circumstances and the adverb is the abbreviated expression. For him adverbs or prepositional phrases do not modify the verb but are an expansion of the sentence ; it will take more than two centuries before this circumstances characteristic is taken into account by linguists. the canonical word order can be changed in russian by the “ fire of imagination” (abbé girard) or “ the free flow of human thoughts” (Lomonosov). But the analysis of n. karamzin’s language shows that the word order in his writings does not depend of ‘ the fire of imagination” or “ the free flow of human thoughts”, but is managed by the desire and the necessity to produce a coherent text, and karamzin had understood that the circumstances could be placed at the beginning of the sentence if the narrator has to indicate the spatial and temporal background of his sentence. Bracquenier Christine. Les « circonstances » chez les grammairiens russes du XVIIIe siècle. In: Revue des études slaves, tome 83, fascicule 2-3, 2012. La lettre et l’esprit : entre langue et culture. Études à la mémoire de Jean Breuillard, sous la direction de Stéphane Viellard, Laure Troubetzkoy et Serge Aslanoff. pp. 675-689.