Search

This paper investigates dynamic dependence between the American Stock Market (S&P 500) and the World Share Market (MSCIW) and examines whether key monetary variables (short and long-term interest rates, interest rate spreads, and exchange rate) explain changes in this relation, during the period January 2000 - June 2016. The methodology includes a Dynamic Copula approach and a Multilayer Perceptron Network. Results suggest that there is interdependence between the American and global stock market and that the dynamic dependence is mainly explained by the short-term interest rate spread, 3-month T-bill’s rate and 3-month London Interbank Offered Rate LIBOR rate. JEL Classification: C45, C58, D53, E49, G15. Resumen: El objetivo de la presente investigación es analizar la dependencia dinámica entre el índice bursátil americano S&P 500 y el índice bursátil mundial (MSCIW), así como, examinar si variables monetarias clave (tasas de interés de corto y largo plazo, diferenciales de tasas de interés y tipo de cambio) explican los cambios en dicha relación de dependencia. El periodo de estudio es de enero de 2000 a junio de 2016, el cual incluye períodos de calma e incertidumbre. La metodología incluye las metodologías de cópula dinámica y red neuronal perceptrón multicapa. Los resultados sugieren que existe un fenómeno de interdependencia entre los mercados bursátiles. Las variaciones en la relación de dependencia se explican por los cambios en el diferencial de tasas de interés de corto plazo (LIBOR 3 meses - T-bill’s 3 meses). Résumé: L’objectif de cette recherche est d’analyser la dépendance dynamique entre l’indice boursier américain S&P 500 et l’indice boursier mondial (MSCIW), ainsi que d’examiner si des variables monétaires clés (taux d’intérêt à court et à long terme, différentiels de taux d’intérêt et taux de change) expliquent les changements dans cette relation de dépendance. La période d’étude s’applique de janvier 2000 à juin 2016, ce qui comprend des périodes de calme et d’incertitude. La méthodologie comprend les méthodologies de la copule dynamique et du réseau de neurones perceptrons multicouches. Les résultats suggèrent qu’il existe un phénomène d’interdépendance entre les marchés boursiers. Les variations du rapport de dépendance s’expliquent par les variations de l’écart de taux d’intérêt à court terme (LIBOR 3 mois - 3 mois de T-bill).

International audience; Geometric uncertainties may jeopardize the performance of parallel manipulators, especially during motion planning. Recent research demonstrated that, during motion planning and due to uncertainties, manipulators may accidentally assume low performance or singular configurations. Thus, reliable motion planning algorithms are required. Very few algorithms were proposed to avoid such problem in parallel manipulators. This paper presents a reliable motion planning technique. First, failure modes are defined. Then, a Monte Carlo simulation is used to provide information on how the manipulator’s uncertainties affect its conditioning. Based on this simulation, probabilities of failure are computed for several manipulator workspace configurations. After that, an artificial neural network metamodel is trained to overcome Monte Carlo’s computational inefficiency on the failure probability estimation. This metamodel is assessed by an iterative strategy that exploits genetic operators to compute optimal trajectories avoiding regions that are considerably affected by uncertainties. Due to its modular methodology, the technique can be easily adapted for different applications. A 3RRR manipulator is used as a case study.

International audience; For over a decade, the multi-agent patrolling task has received a growing attention from the multi-agent community due to its wide range of potential applications. Various algorithms based on reactive and cognitive architectures have been developed. However, the existing patrolling-specific approaches based on deep learning algorithms are still in preliminary stages. In this paper, we propose to integrate a recurrent neural network as part of a strategy for the multi-agent patrolling task. The recurrent neural networks and more specifically the LSTM architecture, as machines to learn temporal series, are well adapted to the multi-agent patrolling problem to the extent that the latter can be viewed as a decision problem over the time. In order to accomplish this study we proposed a formal model of an LSTM-based agent strategy called LSTM Path Maker . The LSTM network is trained over simulation traces of a fully-informed, coordinated and communicating strategy. Then each agent of the new strategy uses its LSTM network to select the next place to visit by feeding it with its current node. Finally this new LSTM-based strategy is evaluated in simulation and compared with two strategy, a cognitive and coordinated one, and a reactive and decentralised one. Preliminary results indicate that the proposed strategy is better than the decentralised one for the criteria of mean interval and quadratic mean interval, and but also close to HPCC for the former.

This paper presents how Artificial Neural Networks (ANN), integrated in a GIS platform, has been used to analyse ancient settlement patterns, in particular fortified villages and medieval settlements in central Italy. The idea at the base of this method is that each settlement system can be seen as an outcome of a social and territorial background. Through the use of different data such as satellite images, historical cartography and archaeological evidences, it is possible to individuate, with the help of ANN methodologies, some of the links between human settlements and the territory. This methodology allows a double approach: firstly the analysis of ancient settlement pattern distribution itself, and secondly the identification of undiscovered archaeological sites. Particular focus will be put on evaluation and quantification of the relationship between late medieval castles (Xth-XIIIth Century A.D.) and environmental variables, and between actual and ancient land use, with the integration of satellite imagery, in particular Landsat, and the use of historical cartography. Cet article montre comment les réseaux de neurones artificiels (ANN), intégrés dans une plateforme GIS, ont été utilisés pour analyser les modèles d’anciens habitats, en particulier des villages fortifiés et des habitats médiévaux en Italie centrale. L’idée de base de cette méthode est que chaque habitat peut être interprété comme une expression du contexte social et territorial. Utilisant des données différentes comme les images satellitaires, la cartographie historique et les évidences archéologiques, il est possible de localiser, grâce à l’utilisation des ANN, quelques-uns des liens présents entre l’habitat humain et le territoire. Cette méthode permet une double approche: l’analyse des habitats mêmes et l’identification de sites inconnus. Une attention particulière a été portée à l’analyse et à la quantification des relations entre les châteaux (Xème-XIIIème siècles) et aux variables territoriales, mais aussi à l’utilisation du sol, à l’intégration d’images satellitaires, en particulier Landsat, et à l’utilisation de la cartographie historique.

A Wireless Sensor Network (WSN) consisting of autonomous sensor nodes can provide a rich stream of sensor data representing physical measurements. A well built Artificial Neural Network (ANN) model needs sufficient training data sources. Facing the limitation of traditional parametric modeling, this paper proposes a standard procedure of combining ANN and WSN sensor data in modeling. Experiments on indoor thermal modeling demonstrated that WSN together with ANN can lead to accurate fine grained indoor thermal models. A new training method "Multi-Pattern Cross Training" (MPCT) is also introduced in this work. This training method makes it possible to merge knowledge from different independent training data sources (patterns) into a single ANN model. Further experiments demonstrated that models trained by MPCT method shew better generalization performance and lower prediction errors in tests using different data sets. Also the MPCT based Neural Network Model has shown advantages in multi-variable Neural Network based Model Predictive Control (NNMPC). Software simulation and application results indicate that MPCT implemented NNMPC outperformed Multiple models based NNMPC in online control efficiency., Face à la limitation de la modélisation paramétrique, nous avons proposé dans cette thèse une procédure standard pour combiner les données reçues a partir de Réseaux de capteurs sans fils (WSN) pour modéliser a l'aide de Réseaux de Neurones Artificiels (ANN). Des expériences sur la modélisation thermique ont permis de démontrer que la combinaison de WSN et d'ANN est capable de produire des modèles thermiques précis. Une nouvelle méthode de formation "Multi-Pattern Cross Training" (MPCT) a également été introduite dans ce travail. Cette méthode permet de fusionner les informations provenant de différentes sources de données d'entraînements indépendants (patterns) en un seul modèle ANN. D'autres expériences ont montré que les modèles formés par la méthode MPCT fournissent une meilleure performance de généralisation et que les erreurs de prévision sont réduites. De plus, le modèle de réseau neuronal basé sur la méthode MPCT a montré des avantages importants dans le multi-variable Model Prédictive Control (MPC). Les simulations numériques indiquent que le MPC basé sur le MPCT a surpassé le MPC multi-modèles au niveau de l'efficacité du contrôle.

A Wireless Sensor Network (WSN) consisting of autonomous sensor nodes can provide a rich stream of sensor data representing physical measurements. A well built Artificial Neural Network (ANN) model needs sufficient training data sources. Facing the limitation of traditional parametric modeling, this paper proposes a standard procedure of combining ANN and WSN sensor data in modeling. Experiments on indoor thermal modeling demonstrated that WSN together with ANN can lead to accurate fine grained indoor thermal models. A new training method "Multi-Pattern Cross Training" (MPCT) is also introduced in this work. This training method makes it possible to merge knowledge from different independent training data sources (patterns) into a single ANN model. Further experiments demonstrated that models trained by MPCT method shew better generalization performance and lower prediction errors in tests using different data sets. Also the MPCT based Neural Network Model has shown advantages in multi-variable Neural Network based Model Predictive Control (NNMPC). Software simulation and application results indicate that MPCT implemented NNMPC outperformed Multiple models based NNMPC in online control efficiency., Face à la limitation de la modélisation paramétrique, nous avons proposé dans cette thèse une procédure standard pour combiner les données reçues a partir de Réseaux de capteurs sans fils (WSN) pour modéliser a l'aide de Réseaux de Neurones Artificiels (ANN). Des expériences sur la modélisation thermique ont permis de démontrer que la combinaison de WSN et d'ANN est capable de produire des modèles thermiques précis. Une nouvelle méthode de formation "Multi-Pattern Cross Training" (MPCT) a également été introduite dans ce travail. Cette méthode permet de fusionner les informations provenant de différentes sources de données d'entraînements indépendants (patterns) en un seul modèle ANN. D'autres expériences ont montré que les modèles formés par la méthode MPCT fournissent une meilleure performance de généralisation et que les erreurs de prévision sont réduites. De plus, le modèle de réseau neuronal basé sur la méthode MPCT a montré des avantages importants dans le multi-variable Model Prédictive Control (MPC). Les simulations numériques indiquent que le MPC basé sur le MPCT a surpassé le MPC multi-modèles au niveau de l'efficacité du contrôle.

This paper shortly presents a project lanced by the Computer Science Department of the "1 Decembrie 1918" University of Alba Iulia. The project is based on the increasing amount and complexity of the earth science data collected by remote sensors. This huge amount of information underscores the need for research into strategies and techniques to facilitate its analysis and understanding. In this project an application of artificial neural networks to human-centered earth science information processing is described.

Endowing an autonomous system like a robot with intelligent behavior is a difficult problem for several reasons. First, behavior is such a wide topic that a general framework paradigm of inspiration must be chosen in order to obtain a consistent model. Such a framework can be for example biological modeling or an artificial intelligence approach. Second, a general framework is not sufficient to determine a fully specified program to be implemented in a robot. Many choices, tuning and tests must be carried out before obtaining a robust system. This paper describes this. First, a biological model is presented, based on the definition of cortex-like automata, representing elementary functions in the perceptive, motor or associative domain. These automata are connected in a network whose architecture, functioning and learning rules are described in a cortical framework. Second, the computational model derived from that biological model is specified. The way units exchange and compute variables through links is explained, with reference to corresponding biological elements. It is then easier to report experiments allowing an autonomous system to learn regularities of a simple environment and to exploit them to satisfy some internal drives. Even if additional biological hints can be added, this model allow us to better understand how a biological model can be implemented and how biological properties can emerge from a distributed set of units.

Colloque avec actes et comité de lecture. internationale.; International audience; Biological inspiration has led to the design of many connectionist models and mechanisms. Among them, memorization mechanisms are of particular importance to endow biologically inspired systems with efficient and consistent adaptive abilities. In this paper, we report recent modelling works of this kind, implementing procedural, episodic and working memories and making them cooperate for autonomous agent navigation.