Search

International audience; In this paper, we address the temporal knapsack problem (TKP). In this generalization of the classical knapsack problem, selected items enter and leave the knapsack at fixed dates. We solve the TKP with a dynamic program of exponential size, which is solved using a method called Successive Sublimation Dynamic Programming (SSDP). This method starts by relaxing a set of constraints from the initial problem, and iteratively reintroduces them when needed. We show that a direct application of SSDP to the temporal knapsack problem does not lead to an effective method, and that several improvements are needed to compete with the best results from the literature.

International audience; In this paper, we study the problem of autonomous and reliable positioning of a camera w.r.t. an object when only this latter is known but not the rest of the scene. We propose to combine the advantages and efficiency of a visual servoing scheme and the generalization ability of a generative adversarial network. The paper describes how to efficiently create a synthetic dataset in order to train a network that predicts an intermediate visual keyframe between two images. Subsequent predictions are used as visual features to autonomously converge towards the desired pose even for large displacements. We show that the proposed method can be used without any prior knowledge on the scene appearance except for the object itself, while being robust to various lighting conditions and specular surfaces. We provide experimental results, both in simulation and using a real service robot platform to validate and evaluate the effectiveness, robustness, and accuracy of our approach.

This paper revisits interval-valued fuzzy regression and proposes a new unified framework to address interval-valued type-1 and type-2 fuzzy regression models. The paper focuses on two main objectives. First, some philosophical and methodological reflections about interval-valued type-1 fuzzy regression (IV-T1FR) and interval-valued type-2 fuzzy regression (IV-T2FR) are discussed and analyzed. These reflections aim at positioning fuzzy regression to avoid misinterpretations that may sometimes lead to erroneous or ambiguous considerations in practical applications. Consequently, the interest, relevance, representativeness and typology of interval-valued fuzzy regression are established. Therefore, IV-T1FR generalizes conventional interval regression (CIR) and increases its specificity. However, if the IV-T1FR can fit fuzzy data, then its formalism is not able to address the uncertainty phenomenon in the IV-T1FS representation. In this context, the IV-T2FR can be regarded as an uncertain IV-T1FR, i.e., a generalization of the IV-T1FR in an uncertain environment. Second, a new unified methodology to address fuzzy regression models using the concepts of gradual intervals (GIs) and thick gradual intervals (TGIs) is proposed. The proposed view allows handling regression models via an extension of the standard interval arithmetic (SIA) – initially proposed for conventional intervals (CIs) – to GIs and TGIs. The originality of the proposed approach resides in the fact that all the CIR methodologies can be extended to the IV-T1FR methods. Furthermore, all the IV-T1FR methodologies can be extended to the IV-T2FR framework. Our view does not depend on the model shape and preserves the flexibility and rigor of SIA computations in the propagation of fuzzy quantities through regression models. The proposed concepts are validated using illustrative examples.

Domain Adaptation aims at benefiting from a labeled dataset drawn from a source distribution to learn a model from examples generated from a different but related target distribution. Creating a domain-invariant representation between the two source and target domains is the most widely technique used. A simple and robust way to perform this task consists in (i) representing the two domains by subspaces described by their respective eigenvectors and (ii) seeking a mapping function which aligns them. In this paper, we propose to use Optimal Transport (OT) and its associated Wassertein distance to perform this alignment. While the idea of using OT in domain adaptation is not new, the original contribution of this paper is two-fold: (i) we derive a generalization bound on the target error involving several Wassertein distances. This prompts us to optimize the ground metric of OT to reduce the target risk; (ii) from this theoretical analysis, we design an algorithm (MLOT) which optimizes a Mahalanobis distance leading to a transportation plan that adapts better. Extensive experiments demonstrate the effectiveness of this original approach.

International audience; In this paper, a novel artificial neural network (ANN) based method dedicated to simultaneously estimating thermal conductivity and thermal diffusivity of CSP (concentrating solar power) plant receiver materials is presented. By monitoring the evolution of these two correlated thermophysical properties during aging cycles, CSP plants' cost efficiency could be maintained. The proposed method is based on the processing of experimental photothermal data using classification and estimation networks. All the networks are feedforward ANN trained with supervised learning algorithms. A pseudo random binary signal (PRBS) is used as excitation and the impact on performance of both the photothermal response length, which is used as model input, and the number of training examples has been evaluated. Of course, the networks' topology has been optimized, allowing the generalization ability to be controlled. Despite the lack of data, the results are promising. Mean relative errors are between 8% and 20%, and the main levers for improvement are identified. In this paper, the study deals with a large range of materials (polymers and metallic alloys).

International audience; This paper deals with the analysis of a class of offline state estimators for LTI discrete-time systems in the presence of an arbitrary measurement noise which can potentially take any value. The considered class of estimators is defined as the solution of an optimization problem involving a performance function which can be interpreted as a generalization of cost functions used in the Maximum Correntropy Criterion. The conclusion of the analysis is that if the system is observable enough, then the considered class of estimators is resilient, which means that the obtained estimation error is independent from the highest values of the measurement noise. In the case of systems with a bounded process noise, the considered class of estimators provides a bounded estimation error under the appropriate conditions despite not being designed for this scenario.

International audience; This paper proposes state and unknown input (UI) observers for linear parameter varying (LPV) systems affected by UI and perturbations in both state and measurement equations. The estimation is done by minimizing the L2 transfer between the perturbations and the state estimation error (H∞-observation). The originality of the paper is to provide a generalization of existing works, specially by relaxing an assumption on systems matrices, widely used for UI decoupling. After giving the main results, some examples illustrate the theoretical contributions and give comparisons to former publications.

International audience; Fuzzy Neural Network (FNN) and Support Vector Machine (SVM) are two prominent and powerful learning models broadly used for classification and regression. FNN has a significant local representation and human reasoning advantage. However, the drawback of such a network is that the focal point of the learning algorithms (e.g., backpropagation) is minimizing empirical risk. In contrary to FNN, SVM emphasizes simultaneously on minimizing empirical and expected risks, which theoretically leads to an excellent generalization performance power. Nevertheless, this depends on determining an adequate kernel function which is consistent with the data properties. Some approaches have been proposed to take advantage of one to deal with disadvantage of the other by combining SVM kernel and FFN. In this paper, we show that a Takagi-Sugeno-Kang (TSK)-type-based fuzzy neural network is, in fact, equivalent to an SVM with an adaptive kernel based on fuzzy rules generated in this FNN. Consequently, it is possible to learn the last layer of the FNN using the concepts of SVM and thus taking the advantage of SVM in generalization. In fact, on the one side, the proposed method is an SVM with an adaptive kernel based on fuzzy rules and on the other side, it is a TSK-FNN with SVM-based learning. Therefore, the FNN with SVM-based learning has the benefit of SVM which is minimizing both learning and testing errors. Besides, such FNN does not have the disadvantages of the SVM which is finding an appropriate kernel and training the kernel parameters. As a matter of fact, the defined kernel in FNN is an adaptive kernel based on data characteristics, which is derived from the fuzzy rules generated by the FNN itself. As a result, it is expected that the results of the proposed method for classification and regression outperform the results of SVM with the conventional kernels and fuzzy neural network trained in the conventional way. This claim is confirmed with the results obtained in this paper.

International audience; Supply chain planning consists of managing a request, a capacity and a performing process. This task is not easy. Indeed, it involves the existence of parameterization rules of the planning process. A parameterization rule describes a causal relationship between management parameters and performance indicators. The academic bibliography brings a significant number of research works to the study of these relationships. In this paper, the research question deals with the potential generalization of the parameterization rules as reviewed from the literature. Based on a dataset of 87 reviewed papers, a set of outcomes are highlighted as a result of a three-step method of analysis. The first step involves the identification of the different parameters and indicators, in addition to the relationships between them. The second step focuses on the examination of the number of papers for each relationship. Finally, the third step is dedicated to the measurement of the generalization degree in the relationship. The results evaluation highlight the contextual nature of the parameterization rules.

We consider the Griffith fracture model in two spatial dimensions, and prove existence of strong minimizers, with closed jump set and continuously differentiable deformation fields. One key ingredient, which is the object of the present paper, is a generalization to the vectorial situation of the decay estimate by De Giorgi, Carriero, and Leaci. This is based on replacing the coarea formula by a method to approximate S B D p functions with small jump set by Sobolev functions, and is restricted to two dimensions. The other two ingredients will appear in companion papers and consist respectively in regularity results for vectorial elliptic problems of the elasticity type and in a method to approximate in energy G S B D p functions by S B V p ones.

This paper discusses a turbulent intermittency model introduced in 1990, the B-model. It was found that the original manuscript which introduced the B-model contained a couple arithmetic errors in the equations. This work goes over corrections to the original equations, and explains where problems arose in the derivations. These corrections cause the results to differ from those in the original manuscript, and these differences are discussed. A generalization of this B-model is then introduced to explore the range of behaviors this style of model provides. To distinguish between the different intermittency models discussed in this paper requires structure function power exponents of order greater than 12. As a source of comparison, data from a flume experiment is introduced, and, with the corrections introduced, this data seems to imply that an intermittency coefficient between 0.17 and 0.2 gives good agreement. Higher quality future measurements of high order moments could help with distinguishing the different intermittency models.

International audience; In this paper, a generalization of the capacitated profitable tour problem is presented. Since recharging time and driver’s lunch breaks are commonly considered as an idle time, the aim is to synchronize those activities choosing restaurants where electric vehicles can be charged. Due to visiting restaurants has an associated cost, the problem can also be seen as a problem with location issues. This variant is pertinent especially in a city logistics context. A mathematical model is proposed, as well as a Branch-and-Price algorithm which is able to solve instances with up to 100 customers and 13 restaurants.

International audience; Tactile maps are essential tools for visually impaired people to comprehend space and to support the simple pedestrian trips made difficult by their disability. Tactile maps are created manually and printed by specialists, and it takes a large amount of time to create a new one, which prevents using them on demand for everyday use. As a consequence, researchers and cartographers try to automate this creation process, but the existing automated derivation processes do not include generalization or advanced stylization steps, which limits their effectiveness. This paper reports first experiments to include such complex automated cartography processes to provide on-demand tactile maps for visually impaired people. These first experiments were more intended to raise real research issues than solve them, and the paper discusses these issues in a research agenda to achieve automatically derived tactile maps.; Les cartes tactiques sont des supports essentiels pour les malvoyants pour appréhender l’espace et pour faciliter les déplacements pédestres rendus difficiles par leur handicap. Les cartes tactiles sont fabriquées manuellement et imprimées par des spécialistes. Il faut donc un temps important pour fabriquer une nouvelle carte ce qui empêche un accès à la demande pour des utilisations quotidiennes. Par conséquent des chercheurs et cartographes essayent d’automatiser le processus de fabrication, mais les processus automatisés actuels de dérivation n’incluent pas les étapes de généralisation et de stylisation avancée ce qui limite leur efficacité. Ce papier présente les premières expériences pour inclure des processus de cartographie automatique complexes pour produire des cartes tactiles à la demande pour les personnes ayant des déficiences visuelles. Ces premières expériences ont davantage pour objectif d’identifier les réelles problématiques de recherche que leur résolution. Le papier discute ces questions dans le cadre d’un agenda de recherche pour parvenir à dériver des cartes tactiles de façon automatique.

Following the publication by Sewell and Jacobson of a Branch and Bound algorithm which has solved all existing benchmarks for the simple assembly line balancing problem of type 1 (SALBP-1) in less than 1 second per problem, this paper investigates the possible existence of harder instances and how they can be designed. SALBP-1 is a generalization of the bin packing problem. We demonstrate how a hard bin packing case can be generated. We prove that, in the worst case, Branch and Bound algorithms with Lower Bounds computed in polynomial time are unable to solve even moderate sized examples of the problem in appropriate time limit. Branch and Bound is an excellent approach based on decomposition and enumeration. Intelligent techniques were elaborated to avoid a complete enumeration. Nevertheless, as demonstrated here, it is always possible to find cases where a Branch and Bound will necessitate a complete enumeration.

Generalized linear models (GLMs) arise in high-dimensional machine learning, statistics, communications and signal processing. In this paper we analyze GLMs when the data matrix is random, as relevant in problems such as compressed sensing, error-correcting codes or benchmark models in neural networks. We evaluate the mutual information (or "free entropy") from which we deduce the Bayes-optimal estimation and generalization errors. Our analysis applies to the high-dimensional limit where both the number of samples and the dimension are large and their ratio is fixed. Non-rigorous predictions for the optimal errors existed for special cases of GLMs, e.g. for the perceptron, in the field of statistical physics based on the so-called replica method. Our present paper rigorously establishes those decades old conjectures and brings forward their algorithmic interpretation in terms of performance of the generalized approximate message-passing algorithm. Furthermore, we tightly characterize, for many learning problems, regions of parameters for which this algorithm achieves the optimal performance, and locate the associated sharp phase transitions separating learnable and non-learnable regions. We believe that this random version of GLMs can serve as a challenging benchmark for multi-purpose algorithms. This paper is divided in two parts that can be read independently: The first part (main part) presents the model and main results, discusses some applications and sketches the main ideas of the proof. The second part (supplementary informations) is much more detailed and provides more examples as well as all the proofs., 101 pages, 5 figures

International audience; For some safety-critical applications, the risk indicator is represented as a time series of estimation errors in the case of a linear model. The safety of the system is compromised if the probability that this risk indicator leaves a given confidence zone at least once during a certain period becomes too important. Sometimes, we are also interested in the calculation of the instantaneous risk probability. The main difficulty is that the Cumulative Distribution Functions (CDFs) (with infinite support) of measurement noise in the above-mentioned linear model are unknown and only their upper and lower bounds are available. The present paper continues the study of previously developed conservative bounds for the above-mentioned risk probabilities as functions of the bounds for the measurement noise CDFs. The original contribution of the present paper consists in the generalization of the previously obtained results to the case of a linear model.

In this paper we present a method to improve and to update the knowledge used for the automation of the generalization of buildings based on agent paradigm. We propose to store 1/ each building decision, 2/ the reason why the decision was taken (the conflicts) 3/ the result of each algorithm (an improvement or not) and 4/ the successful process chain within all trials. At the end, the processes of all buildings are compared in order to identify the weakness (for example the case where a specific algorithm is often used but never succeeds). When a deficiency is identified we introduce new rules and we study the effect of this change on the efficiency of the process. It can be used either to improve existing knowledge or to introduce new rules associate to the use of a new measure or a new algorithm. The first study has been made on building independent generalization to set the learning methodology. We wish now to apply it on more complex cases such as contextual generalization which still needs knowledge improvement.

This paper focuses on the theoretical and numerical development of the J-integral concept for three-dimensional problems. A new integral parameter, developed for the real three-dimensional case, computes the combined energy release rate for an arbitrary crack front. A path-independent integral is verified in both the static and crack propagation cases. Generalization of the J 3 D -integral toward its G θ 3 D implementation form is performed. The efficiency of this proposed integral is then compared with a number of J integrals available in the literature. Lastly, the paper offers a finite element implementation along with its numerical validation.

In this paper, Slepian-Bangs-type formulas for Complex Elliptically Symmetric distributed (CES) data vectors in the presence of model misspecification are provided. The basic Slepian-Bangs (SB) formula has been introduced in the array processing literature as a convenient and compact representation of the Fisher Information Matrix (FIM) for parameter estimation under (parametric) Gaussian data model. Extending recent results on this topic, in this paper, we provide a new generalization of the classical SB formula to parametric estimation problems involving non-Gaussian, heavy-tailed, CES distributed data in the presence of model misspecification. Moreover, we show that our proposed formulas encompass the special cases of the SB formula for CES distributions under perfect model specification, the SB formulas in the presence of misspecified Gaussian models, and the SB formula for the estimation of the scatter matrix of a set of CES distributed data under misspecification of the density generator.

Markov decision processes (MDP) have become one of the standard models for decisiontheoretic planning problems under uncertainty. In its standard form, rewards are assumed to be numerical additive scalars. In this paper, we propose a generalization of this model allowing rewards to be functional. The value of a history is recursively computed by composing the reward functions. We show that several variants of MDPs presented in the literature can be instantiated in this setting. We then identify sufficient conditions on these reward functions for dynamic programming to be valid. We also discuss the infinite horizon case and the case where a maximum operator does not exist. In order to show the potential of our framework, we conclude the paper by presenting several illustrative examples.

In an asynchronous distributed system where any number of processes may crash, a process may have to run solo, computing its local output without receiving any information from other processes. In the basic shared memory system where the processes communicate through atomic read/write registers, at most one process may run solo. This paper introduces the family of d-solo models, where d-processes may concurrently run solo, 1 ≤ d ≤ n (the 1-solo model is the basic read/write model). The paper then studies distributed colorless computations in the d-solo models, where process ids are not used, either in task specifications or during computation. It presents a characterization of the colorless tasks that can be solved in each d-solo model. Colorless tasks include consensus, set agreement and many other previously studied tasks. It shows that colorless algorithms have limited computational power for solving tasks, only when d > 1 . When d = 1 , colorless algorithms can solve the same tasks as algorithms that may use ids. It is well-known that, while consensus is not wait-free solvable in a model where at most one process may run solo, ϵ-approximate agreement is solvable. In a d-solo model, the fundamental solvable task is ( d , ϵ ) -solo approximate agreement, a generalization of ϵ-approximate agreement. Indeed, ( d , ϵ ) -solo approximate agreement can be solved in the d-solo model, but not in the ( d + 1 ) -solo model. Finally, the paper studies a link between the solvability of d-set agreement and ( d , ϵ ) -solo approximate agreement in asynchronous wait-free message-passing systems, which provides an insight on the “maximal partitioning” allowed to solve an approximate agreement task.

The proposed approach solves the Multi-Depot Open VRP.An uniform view of ejection chains is adopted for all neighborhoods.Multiple Neighborhood Search (MNS) and Tabu Search (TS) are combined.MNS-TS outperforms the state-of-the-art methods. In this paper, we address the Multi-Depot Open Vehicle Routing Problem (MDOVRP), which is a generalization of the Capacitated Vehicle Routing Problem (CVRP) where vehicles start from different depots, visit customers, deliver goods and are not required to return to the depot at the end of their routes. The goal of this paper is twofold. First, we have developed a general Multiple Neighborhood Search hybridized with a Tabu Search (MNS-TS) strategy which is proved to be efficient and second, we have settled an unified view of ejection chains to be able to handle several neighborhoods in a simple manner. The neighborhoods in the proposed MNS-TS are generated from path moves and ejection chains. The numerical and statistical tests carried out over OVRP and MDOVRP problem instances from the literature show that MNS-TS outperforms the state-of-the-art methods.

International audience; This paper deals with a novel generalization of classical blind source separation (BSS) in two directions. First, relaxing the constraint that the latent sources must be statistically independent. This generalization is well-known and sometimes termed independent subspace analysis (ISA). Second, jointly analyzing several ISA problems, where the link is due to statistical dependence among corresponding sources in different mixtures. When the data are one-dimensional, i.e., multiple classical BSS problems, this model, known as independent vector analysis (IVA), has already been studied. In this paper, we combine IVA with ISA and term this new model joint independent subspace analysis (JISA). We provide full performance analysis of JISA, including closed-form expressions for minimal mean square error (MSE), Fisher information and Cramér-Rao lower bound, in the separation of Gaussian data. The derived MSE applies also for non-Gaussian data, when only second-order statistics are used. We generalize previously known results on IVA, including its ability to uniquely resolve instantaneous mixtures of real Gaussian stationary data, and having the same arbitrary permutation at all mixtures. Numerical experiments validate our theoretical results and show the gain with respect to two competing approaches that either use a finer block partition or a different norm.

International audience; The problem of partitioning a square into zones of prescribed areas arises when partitioning matrices for dense linear algebra kernels onto a set of heterogeneous processors, and several approximation algorithms have been proposed for that problem. In this paper, we address the natural generalization of this problem in dimension 3: partition a cuboid in a set of zones of prescribed volumes (which represent the amount of computations to perform), while minimizing the surface of the boundaries between zones (which represent the data transfers involved). This problem naturally arises in the context of matrix multiplication, and can be seen as a heterogeneous generalization of 2.5D approaches that have been proposed in this context. The contributions of this paper are twofold. We prove the NP-completeness of the general problem, and we propose a 5/6^(2/3) (~1.51) approximation algorithm for cube-partitioning. This is the first known approximation result for this 3D partitioning problem.

International audience; This paper addresses the generalisation of stationary Hawkes processes in order to allow for a time-evolving second-order analysis. Motivated by the concept of locally stationary autoregressive processes, we apply however inherently different techniques to describe the time-varying dynamics of self-exciting point processes. In particular we derive a stationary approximation of the Laplace transform of a locally stationary Hawkes process. This allows us to define a local intensity function and a local Bartlett spectrum which can be used to compute approximations of first and second order moments of the process. We complete the paper by some insightful simulation studies.

Characterizing fitness landscapes associated with polymorphic adaptive traits enables investigation of mechanisms allowing transitions between fitness peaks. Here, we explore how natural selection can promote genetic mechanisms preventing heterozygous phenotypes from falling into non-adaptive valleys. Polymorphic mimicry is an ideal system to investigate such fitness landscapes, because the direction of selection acting on complex mimetic colour patterns can be predicted by the local mimetic community composition. Using more than 5000 artificial butterflies displaying colour patterns exhibited by the polymorphic Müllerian mimicHeliconius numata, we directly tested the role of wild predators in shaping fitness landscapes. We compared predation rates on mimetic phenotypes (homozygotes at the supergene controlling colour pattern), intermediate phenotypes (heterozygotes), exotic morphs (absent from the local community) and palatable cryptic phenotypes. Exotic morphs were significantly more attacked than local morphs, highlighting predators' discriminatory capacities. Overall, intermediates were attacked twice as much as local homozygotes, suggesting the existence of deep fitness valleys promoting strict dominance and reduced recombination between supergene alleles. By including information on predators' colour perception, we also showed that protection on intermediates strongly depends on their phenotypic similarity to homozygous phenotypes and that ridges exist between similar phenotypes, which may facilitate divergence in colour patterns.

Oblivious Transfer $$\mathsf {OT} $$ protocols were introduced in the seminal paper of Rabin, and allow a user to retrieve a given number of lines usually one in a database, without revealing which ones to the server. The server is ensured that only this given number of lines can be accessed per interaction, and so the others are protected; while the user is ensured that the server does not learn the numbers of the lines required. This primitive has a huge interest in practice, for example in secure multi-party computation, and directly echoes to Symmetrically Private Information Retrieval SPIR. Recent Oblivious Transfer instantiations secure in the UC framework suffer from a drastic fallback. After the first query, there is no improvement on the global scheme complexity and so subsequent queries each have a global complexity of $$\mathcal {O}|DB|$$ meaning that there is no gain compared to running completely independent queries. In this paper, we propose a new protocol solving this issue, and allowing to have subsequent queries with a complexity of $$\mathcal {O}\log |DB|$$ while keeping round optimality, and prove the protocol security in the UC framework with adaptive corruptions and reliable erasures. As a second contribution, we show that the techniques we use for Oblivious Transfer can be generalized to a new framework we call Oblivious Language-Based Envelope$$\mathsf {OLBE}$$. It is of practical interest since it seems more and more unrealistic to consider a database with uncontrolled access in access control scenarios. Our approach generalizes Oblivious Signature-Based Envelope, to handle more expressive credentials and requests from the user. Naturally, $$\mathsf {OLBE}$$ encompasses both $$\mathsf {OT} $$ and $$\mathsf {OSBE}$$, but it also allows to achieve Oblivious Transfer with fine grain access over each line. For example, a user can access a line if and only if he possesses a certificate granting him access to such line. We show how to generically and efficiently instantiate such primitive, and prove them secure in the Universal Composability framework, with adaptive corruptions assuming reliable erasures. We provide the new UC ideal functionalities when needed, or we show that the existing ones fit in our new framework. The security of such designs allows to preserve both the secrecy of the database values and the user credentials. This symmetry allows to view our new approach as a generalization of the notion of Symmetrically PIR.

International audience; In this paper, we propose an interactive approach to determine a compromise solution in the multicriteria spanning tree problem. We assume that the Decision Maker’s preferences over spanning trees can be represented by a weighted sum of criteria but that weights are imprecisely known. In the first part of the paper, we propose a generalization of Prim’s algorithm to determine the set of possibly optimal tradeoffs. In the second part, we propose an incremental weight elicitation method to reduce the set of feasible weights so as to identify a necessary optimal tradeoff. Numerical tests are given to demonstrate the practical feasibility of the approach.

International audience; The paper deals with the valued comparison of intervals for decision making. Interval orders are classical preference structures where the comparison of intervals is done in an ordinal way. In this paper we focus on valued comparison where more information, especially the distance between end-points of intervals, is used in order to have more sophisticated preference structures. The generalization of an interval order as a valued structure requires the choice of de Morgan triplets. We propose a valued outranking relation for interval comparison and show that it satisfies different definitions of valued interval orders using different de Morgan triplet. The decomposition of our outranking relation into preference and indifference provides a valued preference structure where the preference is T-transitive and monotone.

International audience; —Performances of data-driven prognostics approaches are closely dependent on form, and trend of extracted features. Indeed, features that clearly reflect the machine degradation, should lead to accurate prognostics, which is the global objective of the paper. This paper contributes a new approach for features extraction / selection: the extraction is based on trigonomet-ric functions and cumulative transformation, and the selection is performed by evaluating feature fitness using monotonicity and trendability characteristics. The proposition is applied to time-frequency analysis of non-stationary signals using Discrete Wavelet Transform. The main idea is to map raw vibration data into monotonic features with early trends, which can be easily predicted. To show that, selected features are used to build a model with a data-driven approach namely, the Summation Wavelet-Extreme Learning Machine, that enables a good balance between model accuracy and complexity. For validation and generalization purpose, vibration data from two real applications of Prognostics and Health Management challenges are used: 1) cutting tools from Computer Numerical Control (CNC) machine (2010), and 2) bearings from platform PRONOSTIA (2012). Performances of the proposed approach are thoroughly compared with the classical approach by performing: feature fitness analysis, cutting tool wear " estimation " and bearings " long-term predictions " tasks, which validates the proposition.

International audience; Markov decision processes (MDP) have become one of the standard models for decision-theoretic planning problems under uncertainty. In its standard form, rewards are assumed to be numerical additive scalars. In this paper, we propose a generalization of this model allowing rewards to be functional. The value of a history is recursively computed by composing the reward functions. We show that several variants of MDPs presented in the literature can be instantiated in this setting. We then identify sufficient conditions on these reward functions for dynamic programming to be valid. In order to show the potential of our framework, we conclude the paper by presenting several illustrative examples.

International audience; Recent technologies allowed a major growth of geographical datasets with different levels of detail, different point of views, and different specifications, but on the same geographical extent. These datasets need to be integrated in order to benefit from their diversity. Conflation is one of the solutions to provide integration. Conflation aims at combining data that represent same entities from several datasets, into a richer new dataset. This paper proposes a framework that provides a geometrical conflation that preserves the characteristic shapes of geographic data. The framework is based on least squares adjustment, inspired from least squares based generalization techniques. It does not require very precise pre-matching, which is interesting as automatic matching remains a challenging task. Several constraints are proposed to preserve different kind of shape and relations between features, while conflating data. The framework is applied to a real land use parcels conflation problem with excellent results. The least squares based conflation is evaluated, notably with comparisons with existing techniques like rubber sheeting. The paper also describes how the framework can be extended to other geometrical optimization problems.

International audience; This paper considers the modelling of the process of Corrective and condition-based Preventive Maintenance, for complex repairable systems. In order to take into account the dependency between both types of maintenance and the possibility of imperfect maintenance, Generalized Competing Risks models have been introduced in "Doyen and Gaudoin (J Appl Probab 43:825-839, 2006)". In this paper, we study two classes of these models, the Generalized Random Sign and Generalized Alert Delay models. A Generalized Competing Risks model can be built as a generalization of a particular Usual Competing Risks model, either by using a virtual age framework or not. The models properties are studied and their parameterizations are discussed. Finally, simulation results and an application to real data are presented.

International audience; Since the 1970s, extensive research has been conducted to improve the performance of MRP systems (Material Requirement Planning). We identified 86 papers in this domain. These papers examine the influence of configuration on performance measurement indicators in defined operating and instability conditions. These studies, in principle, provide production planners with a decision-making aid. The initial objective of this paper is to measure and characterize the accumulated knowledge in the 86 papers identified. The potential generalization of the knowledge gained is subsequently discussed. We propose a method of analysis of this knowledge.

International audience; In this paper, we propose to explain Discounted Cumulative Gain (DCG) as the expectation of the total utility collected by a user given a generative probabilistic model on how users browse the result page ranking list of a search engine. We contrast this with a generalization of Average Precision, pAP, that has been defined in Dupret and Piwowarski (2010). In both cases, user decision models coupled with Web search logs allow to estimate some parameters that are usually left to the designer of a metric. In this paper, we compare the user models for DCG and pAP at the interpretation and experimental level.DCG and AP are metrics computed before a ranking function is exposed to users and as such, their role is to predict the function performance. In counterpart to prognostic metric, a diagnostic metric is computed after observing the user interactions with the result list. A commonly used diagnostic metric is the clickthrough rate at position 1, for example. In this work we show that the same user model developed for DCG can be used to derive a diagnostic version of this metric. The same hold for pAP and any metric with a proper user model.We show that not only does this diagnostic view provide new information, it also allows to define a new criterion for assessing a metric. In previous works based on user decision modeling, the performance of different metrics were compared indirectly in terms of the ability of the associated user model to predict future user actions. Here we propose a new and more direct criterion based on the ability of the prognostic version of the metric to predict the diagnostic performance.

In this paper we explore the connections between two classes of polyominoes, namely the permutominoes and the pseudo-square polyominoes. A permutomino is a polyomino uniquely determined by a pair of permutations. Permutominoes, and in particular convex permutominoes, have been considered in various kinds of problems such as: enumeration, tomographical reconstruction, and algebraic characterization. On the other hand, pseudo-square polyominoes are a class of polyominoes tiling the plane by translation. The characterization of such objects has been given by Beauquier and Nivat, who proved that a polyomino tiles the plane by translation if and only if it is a pseudo-square or a pseudo-hexagon. In particular, a polyomino is pseudo-square if its boundary word may be factorized as [email protected]^[email protected]^, where [email protected]^ denotes the path X traveled in the opposite direction. In this paper we relate the two concepts by considering the pseudo-square polyominoes which are also convex permutominoes. By using the Beauquier-Nivat characterization we provide some geometrical and combinatorial properties of such objects, and we show for any fixed X, each word Y such that [email protected]^[email protected]^ is pseudo-square is prefix of a unique infinite word Y"~ with period 4|X|"N|X|"E. Also, we show that [email protected]^[email protected]^ are centrosymmetric, i.e. they are fixed by rotation of angle @p. The proof of this fact is based on the concept of pseudoperiods, a natural generalization of periods.

International audience; In this paper, we aim at modeling and analyzing the regulation processes in multi-cellular biological systems, in particular tissues. The modeling frame- work is a generalization of several existing formalisms. In particular, it can be seen as an extension of logical regulatory networks ('a la Thomas) with information about cells physical state and environment, e.g., their spatial re- lationships. The resulting formalisms, called integrated regulatory networks (IRNs) is equipped with a transition systems semantics that preserves the possibility of an enumerative and exhaustive state space exploration. This paper presents the modeling framework, its semantics, as well as a prototype implementation that allowed preliminary experiments on some applications related to biology.

International audience; Engineering sciences and applications of mathematics show unambiguously that positive semidefiniteness of matrices is the most important generalization of non-negative real num- bers. This notion of non-negativity for matrices has been well-studied in the literature; it has been the subject of review papers and entire chapters of books. This paper reviews some of the nice, useful properties of positive (semi)definite matrices, and insists in particular on (i) characterizations of positive (semi)definiteness and (ii) the geometrical properties of the set of positive semidefinite matrices. Some properties that turn out to be less well-known have here a special treatment. The use of these properties in optimization, as well as various references to applications, are spread all the way through. The "raison d'être" of this paper is essentially pedagogical; it adopts the viewpoint of variational analysis, shedding new light on the topic. Important, fruitful, and subtle, the positive semidefinite world is a good place to start with this domain of applied mathematics.

International audience; This paper concerns the study of the correlation measures of finite binary sequences, more particularly the dependence of correlation measures of even order and correlation measures of odd order. These results generalize previous results due to Gyarmati (2005)171 and to Anantharam (2008) [3] and provide a partial answer to a conjecture due to Mauduit (2003)[12]. The last part of the paper concerns the generalization of this study to the case of finite binary n-dimensional lattices.

International audience; An AMONG constraint holds if the number of variables that belong to a given value domain is between given bounds. This paper focuses on the case where the variable and value domains are intervals. We investigate the conjunction of AMONG constraints of this type. We prove that checking for satisfiability -- and thus, enforcing bound consistency -- can be done in polynomial time. The proof is based on a specific decomposition that can be used as such to filter inconsistent bounds from the variable domains. We show that this decomposition is incomparable with the natural conjunction of \textsc{Among} constraints, and that both decompositions do not ensure bound consistency. Still, experiments on randomly generated instances reveal the benefits of this new decomposition in practice. This paper also introduces a generalization of this problem to several dimensions and shows that satisfiability is NP-complete in the multi-dimensional case.

International audience; This paper introduces Pairwise Constrained Component Analysis (PCCA), a new algorithm for learning distance metrics from sparse pairwise similarity/dissimilarity constraints in high dimensional input space, problem for which most existing distance metric learning approaches are not adapted. PCCA learns a projection into a low-dimensional space where the distance between pairs of data points respects the desired constraints, exhibiting good generalization properties in presence of high dimensional data. The paper also shows how to efficiently kernelize the approach. PCCA is experimentally validated on two challenging vision tasks, face verification and person re-identification, for which we obtain state-of-the-art results.

In this paper we classify all extremal and $s$-extremal binary self-dual codes of length 38. There are exactly 2744 extremal $[38,19,8]$ self-dual codes, two $s$-extremal $[38,19,6]$ codes, and 1730 $s$-extremal $[38,19,8]$ codes. We obtain our results from the use of a recursive algorithm used in the recent classification of all extremal self-dual codes of length 36, and from a generalization of this recursive algorithm for the shadow. The classification of $s$-extremal $[38,19,6]$ codes permits to achieve the classification of all $s$-extremal codes with d=6., revised version - paper submitted (4/4/2011) to IEEE trans. Information and accepted 20/10/2011

In this paper we investigate operators unitarily equivalent to trun- cated Toeplitz operators. We show that this class contains certain sums of tensor products of truncated Toeplitz operators. In particular, it contains arbitrary inflations of truncated Toeplitz operators; this answers a question posed in (4). model spaces, are a generalization of the operators associated with Toeplitz matri- ces. They are introduced and discussed in great detail in a recent survey paper by Sarason (10). Some special cases have appeared long ago in the literature: the model operators for contractions with defect number one as well as their commutant are truncated Toeplitz operators with analytic symbols (see, for instance, (9, 11, 8)). This is a new area of study, and many simple questions remain open. The basic reference for this subject is (10), subsequent work is done in (4, 6, 3, 2).

International audience; At ECOOP 2010 in Maribor, Slovenia, the two workshops MASPEGHI (MechAnisms for SPEcialization, Generalization and inHerItance) and ICOOOLPS (Implementation, Compilation, Optimization of Object-Oriented Languages, Programs and Systems) were combined because both were rather small and shared common concerns, their topic areas being strongly related. Six papers had been accepted to MASPEGHI, but only five were presented because the authors of one paper could not attend the conference and workshop. Three papers had been accepted to ICOOOLPS, and all were also presented. The workshop authors were later asked to submit extended versions of their papers for possible publication in this special section. We received two extended papers from ICOOOLPS and one from MASPEGHI. They were carefully reviewed by three reviewers each, and then revised by the authors according to the reviewers' comments. In our opinion, all revised papers were interesting, of high quality and significantly extended from the workshop versions. One of them, however, needed more work from its authors, and they could not complete it within a reasonable time. As a consequence, only two extended, reviewed and revised papers are now published in this special section.

International audience; In many real-life applications, the available source training information is either too small or not representative enough of the underlying target test problem. In the past few years, a new line of machine learning research has been developed to overcome such awkward situations, called Domain Adaptation (DA), giving rise to many adaptation algorithms and theoretical results in the form of generalization bounds. In this paper, a novel contribution is proposed in the form of a DA algorithm dealing with string-structured data, inspired from the DA support vector machine (SVM) technique introduced in [Bruzzone et al, PAMI 2010]. To ensure the convergence of SVM-based learning, the similarity functions involved in the process must be valid kernels, i.e. positive semi-definite (PSD) and symmetric. However, in the string-based context that we are considering in this paper, this condition is often not satisfied. Indeed, it has been proven that most string similarity functions based on the edit distance are not PSD. To overcome this drawback, we make use in this paper of the new theory of learning with good similarity functions introduced by Balcan et al., which (i) does not require the use of a valid kernel to learn well and (ii) allows us to induce sparser models. We take advantage of this theoretical framework to propose a new DA algorithm using good edit similarity functions. Using a suitable string-representation of handwritten digits, we show that are our new algorithm is very efficient to deal with the scaling and rotation problems usually encountered in image classification.

This paper introduces a generalization of the notion of Nash equilibrium (NE), namely quantal response equilibrium (QRE) in the context of wireless communications. In the QRE, radio devices choose their transmit/receive configuration taking into account that the estimation of their own performance contains a noise component. Here, it is shown that the notion of QRE neatly models decentralized self-configuring networks (DCSN) where feedback messages are impaired by quantization noise or decoding errors. The main contribution of the paper is twofold. First, we show that under the presence of noise in the estimation expected utility, the notion of NE no longer holds, as players cannot be considered rational. Second, we introduce a learning technique that converges to a QRE in a fully decentralized fashion. We present numerical results in the context of a channel selection problem in a parallel multiple access channel in order to illustrate our theoretical results.

Special issue on Advances in Angle-of-Arrival and Multidimensional Signal Processing for Localization and Communications; International audience; The Statistical Resolution Limit (SRL), which is defined as the minimal separation between parameters to allow a correct resolvability, is an important statistical tool to quantify the ultimate performance for parametric estimation problems. In this paper we generalize the concept of the SRL to the Multidimensional SRL (MSRL) applied to the multidimensional harmonic retrieval model. In this paper, we derive the SRL for the so-called multidimensional harmonic retrieval model by using a generalization of the previously introduced SRL concepts that we call Multidimensional SRL (MSRL). We first derive the MSRL using an hypothesis test approach. This statistical test is shown to be asymptotically an uniformly most powerful test which is the strongest optimality statement that one could expect to obtain. Second, we link the proposed asymptotic MSRL based on the hypothesis test approach to a new extension of the SRL based on the Cramér-Rao Bound approach. Thus, a closed-form expression of the asymptotic MSRL is given and analyzed in the framework of the multidimensional harmonic retrieval model. Particularly, it is proved that the optimal MSRL is obtained for equi-powered sources and/or an equi-distributed number of sensors on each multi-way array.

We consider a queueing network of symmetric G-queues with customers and signals. Since the seminal papers by Gelenbe in the early 1990s (Gelenbe, E. (1991) Product-form queueing networks with negative and positive customers. J. Appl. Probab., 28, 656–663; Gelenbe, E. (1993); G-networks with instantaneous customer movement. J. Appl. Probab., 30, 742–748; Gelenbe, E. (1993) G-Networks with signals and batch removal. Probab. Eng. Inform. Sci., 7, 335–342), generalized networks of queues have received considerable attention. But most papers assume to obtain a product form such that the service times follow exponential distributions. Here, we propose a new generalization of this model with phase-type service times. We also assume a new type of signal. When a signal enters a queue, it changes the phase of the customer in service if there is any. As usual, after completion of its service, a customer moves to another queue and may become a signal. We prove that the steady-state distribution for such a network of queues has a product form solution.

International audience; Hierarchies are crucial for analysis in data warehouses. But they can hardly be defined on measure attributes. In this paper, we tackle this issue and we show that measure generalizations often depend on a context. For instance, a given blood pressure can be either low, normal or high regarding not only the collected measure but also characteristics of the patient such as the age. The contribution of this paper is threefold. (1) Thanks to an external database storing the expert knowledge, we propose an effective solution for considering these hierarchies. (2) In order to efficiently manage this knowledge, a Rich Internet Application is developed. (3) Finally, in order to provide a flexible analysis, query rewriting module is proposed. Thus, it is possible to answer queries such as: "Who had a low blood pressure last night?''

A fruitful analogy between possibility theory and formal concept analysis has recently contributed to show the interest of introducing new operators in this latter setting. In particular, another Galois connection, distinct from the classical one that defines formal concepts, has been laid bare which allows for the decomposition of a formal context into sub-contexts when possible. This paper pursues a similar investigation by considering pattern structures which are known to offer a generalization of formal concept analysis. The new operators as well as the other Galois connection are introduced in this framework, where an object is associated to a structured description rather than just to its set of properties. The description may take many different forms. In this paper, we more particularly focus on two important particular cases, namely ordered lists of intervals, and propositional knowledge bases, which both allow for incomplete descriptions. They are then extended to fuzzy and uncertain descriptions by introducing fuzzy intervals and possibilistic logic bases respectively in these two settings. © 2010 Springer-Verlag Berlin Heidelberg.