Dissertations / Theses on the topic 'Spatial Intelligence'

La thèse est organisée en deux parties. Une première partie présente la littérature consacrée à l'étude de la cognition spatiale. Cette présentation s'inscrit dans une approche interdisciplinaire (psychologie, neurophysiologie, sciences de l'information, éthologie, philosophie). Il est notamment analysé comment les différences interindividuelles en cognition spatiale sont décrites par ces différentes disciplines. Une attention particulière sera donnée à des différences qualitatives de traitement, c'est-à-dire aux stratégies mises en oeuvre pour réaliser des traitements visuo-spatiaux. On analysera également les différences dans la qualité des représentations construites à partir d'une exploration d'un espace tridimensionnel. Il sera essayé de dresser un cadre interprétatif intégratif des différences constatées sur la base de la théorie neurologiquement plausible de Kosslyn introduisant la distinction entre un subsystème catégoriel et métrique pour l'encodage des relations spatiales. L'hypothèse est émise que l'efficacité de fonctionnement de ces subsystèmes peut constituer un élément explicatif important pour les différences interindividuelles en cognition spatiale. Dans la deuxième partie, on présentera 4 expériences impliquant la mesure de la qualité de l'encodage spatial à l'aide de dispositifs sur support informatique. Un de ces dispositifs est plus spécialement destiné à évaluer la précision de l'encodage spatial et montre la stabilité des différences interindividuelles constatées, ainsi que la relation entre précision d'encodage et performance à des épreuves de type papier-crayon. Il s'avère que cette dernière relation est importante, surtout en ce qui concerne les facteurs spatial et général. Une épreuve d'exploration d'un espace virtuel en 3D permet en plus d'évaluer l'impact de la précision de l'encodage spatial sur l'efficacité de déplacement. Les résultats sont interprétés dans le cadre neurologiquement plausible de la théorie de Kosslyn
The dissertation is organized in two parts. The first part presents the literature concerning the study of spatial cognition. This representation adopts an interdisciplinary approach (psychology, neurophysiology, information sciences, ethology, philosophy). Of special interest are interindividual differences in spatial cognition as described by these disciplines. Particular attention is payed to qualitative processing differences, i. E. Different strategies for processing of visuo-spatial information. Individual differences in the quality of representations constructed from navigation in a threedimensional space are also analysed. The attempt is made to create an integrative framework of interpretation for the described differences on the basis of the neurologically plausible theory of Kosslyn introducing the distinction between categorical and metric spatial relations encoding. The hypothesis is made that the efficiency of these subsystems may represent an important element in the explanation of interindividual differences in spatial cognition. In the second part, 4 experiences are presented which imply the measurement of the quality of spatial relations encoding with computer-assisted tests. One of these tests measures specifically the precision of spatial relations encoding and shows stability of individual differences, as well as the relation between spatial relations encoding precision and performance on paper and pencil tests. This relation seems important, especially with the spatial and general factors of intelligence. A navigation test in a virtual environment furthermore permits to evaluate the influence of spatial relations encoding precision on wayfinding performance. Results are interpreted in the framework of Kosslyn's neurologically plausible theory of visuo-spatial information processing

This thesis investigates the conceptual and methodological issues for the development of Intelligent Spatial Decision Support Systems (ISDSS). These are spatial decision support systems (SDSS) integrating intelligent systems techniques (Genetic Algorithms, Neural Networks, Expert Systems, Fuzzy Logic and Nonlinear methods) with traditional modelling and statistical methods for the analysis of spatial problems. The principal aim of this work is to verify the feasibility of heterogeneous systems for spatial decision support derived from a combination of traditional numerical techniques and intelligent techniques in order to provide superior performance and functionality to that achieved through the use of traditional methods alone. This thesis is composed of four distinct sections: (i) a taxonomy covering the employment of intelligent systems techniques in specific applications of geographical information systems and SDSS; (ii) the development of a prototype ISDSS; (iii) application of the prototype ISDSS to modelling the spatiotemporal dynamics of high technology industry in the South-East of England; and (iv) the development of ISDSS architectures utilising interapplication communication techniques. Existing approaches for implementing modelling tools within SDSS and GIS generally fall into one of two schemes - loose coupling or tight coupling - both of which involve a tradeoff between generality and speed of data interchange. In addition, these schemes offer little use of distributed processing resources. A prototype ISDSS was developed in collaboration with KPMG Peat Marwick's High Technology Practice as a general purpose spatiotemporal analysis tool with particular regard to modelling high technology industry. The GeoAnalyser system furnishes the user with animation and time plotting tools for observing spatiotemporal dynamics; such tools are typically not found in existing SDSS or GIS. Furthermore, GeoAnalyser employs the client/server model of distributed computing to link the front end client application with the back end modelling component contained within the server application. GeoAnalyser demonstrates a hybrid approach to spatial problem solving - the application utilises a nonlinear model for the temporal evolution of spatial variables and a genetic algorithm for calibrating the model in order to establish a good fit for the dataset under investigation. Several novel architectures are proposed for ISDSS based on existing distributed systems technologies. These architectures are assessed in terms of user interface, data and functional integration. Implementation issues are also discussed. The research contributions of this work are four-fold: (i) it lays the foundation for ISDSS as a distinct type of system for spatial decision support by examining the user interface, performance and methodological requirements of such systems; (ii) it explores a new approach for linking modelling techniques and SDSS; (iii) it investigates the possibility of modelling high technology industry; and (iv) it details novel architectures for ISDSS based on distributed systems.

Thesis (M.S. in Information Technology Management) Naval Postgraduate School, Sept. 2000.
"September 2000." Thesis advisor(s): Darken, Rudolph P.; Brady, Terrance C. Includes bibliographical references (p. 77-78). Also available online.

Thesis (M.Eng.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2002.
Includes bibliographical references (leaves 38-40).
by Thomas A. Stocky.
M.Eng.

This thesis presents a formal study of the long-term spatial load forecasting problem: given small area based electric load history of the service territory, current and future land use information, return forecast load of the next 20 years. A hierarchical S-curve trending method is developed to conduct the basic forecast. Due to uncertainties of the electric load data, the results from the computerized program may conflict with the nature of the load growth. Sometimes, the computerized program is not aware of the local development because the land use data lacks such information. A human-machine co-construct intelligence framework is proposed to improve the robustness and reasonability of the purely computerized load forecasting program. The proposed algorithm has been implemented and applied to several utility companies to forecast the long-term electric load growth in the service territory and to get satisfying results.

La reconstruction de bâtiments à partir de photographies aériennes et d’autres données spatiales urbaines multi-sources est une tâche qui utilise une multitude de méthodes automatisées et semi-automatisées allant des processus ponctuels au traitement classique des images et au balayage laser. Dans cette thèse, un système de relaxation itératif est développé sur la base de l'examen du contexte local de chaque bord en fonction de multiples sources d'entrée spatiales (masques optiques, d'élévation, d'ombre et de feuillage ainsi que d'autres données prétraitées, décrites au chapitre 6). Toutes ces données multisource et multirésolution sont fusionnées de manière à extraire les segments de ligne probables ou les arêtes correspondant aux limites des bâtiments. Deux nouveaux sous-systèmes ont également été développés dans cette thèse. Ils ont été conçus dans le but de fournir des informations supplémentaires, plus fiables, sur les contours des bâtiments dans une future version du système de relaxation proposé. La première est une méthode de réseau de neurones à convolution profonde (CNN) pour la détection de frontières de construction. Le réseau est notamment basé sur le modèle SRCNN (Dong C. L., 2015) de super-résolution à la pointe de la technologie. Il accepte des photographies aériennes illustrant des données de zones urbaines densément peuplées ainsi que leurs cartes d'altitude numériques (DEM) correspondantes. La formation utilise trois variantes de cet ensemble de données urbaines et vise à détecter les contours des bâtiments grâce à une nouvelle cartographie hétéroassociative super-résolue. Une autre innovation de cette approche est la conception d'une couche de perte personnalisée modifiée appelée Top-N. Dans cette variante, l'erreur quadratique moyenne (MSE) entre l'image de sortie reconstruite et l'image de vérité de sol (GT) fournie des contours de bâtiment est calculée sur les 2N pixels de l'image avec les valeurs les plus élevées. En supposant que la plupart des N pixels de contour de l’image GT figurent également dans les 2N pixels supérieurs de la reconstruction, cette modification équilibre les deux catégories de pixels et améliore le comportement de généralisation du modèle CNN. Les expériences ont montré que la fonction de coût Top-N offre des gains de performance par rapport à une MSE standard. Une amélioration supplémentaire de la capacité de généralisation du réseau est obtenue en utilisant le décrochage. Le deuxième sous-système est un réseau de convolution profonde à super-résolution, qui effectue un mappage associatif à entrée améliorée entre les images d'entrée à basse résolution et à haute résolution. Ce réseau a été formé aux données d’altitude à basse résolution et aux photographies urbaines optiques à haute résolution correspondantes. Une telle différence de résolution entre les images optiques / satellites optiques et les données d'élévation est souvent le cas dans les applications du monde réel
Building reconstruction from aerial photographs and other multi-source urban spatial data is a task endeavored using a plethora of automated and semi-automated methods ranging from point processes, classic image processing and laser scanning. In this thesis, an iterative relaxation system is developed based on the examination of the local context of each edge according to multiple spatial input sources (optical, elevation, shadow & foliage masks as well as other pre-processed data as elaborated in Chapter 6). All these multisource and multiresolution data are fused so that probable line segments or edges are extracted that correspond to prominent building boundaries.Two novel sub-systems have also been developed in this thesis. They were designed with the purpose to provide additional, more reliable, information regarding building contours in a future version of the proposed relaxation system. The first is a deep convolutional neural network (CNN) method for the detection of building borders. In particular, the network is based on the state of the art super-resolution model SRCNN (Dong C. L., 2015). It accepts aerial photographs depicting densely populated urban area data as well as their corresponding digital elevation maps (DEM). Training is performed using three variations of this urban data set and aims at detecting building contours through a novel super-resolved heteroassociative mapping. Another innovation of this approach is the design of a modified custom loss layer named Top-N. In this variation, the mean square error (MSE) between the reconstructed output image and the provided ground truth (GT) image of building contours is computed on the 2N image pixels with highest values . Assuming that most of the N contour pixels of the GT image are also in the top 2N pixels of the re-construction, this modification balances the two pixel categories and improves the generalization behavior of the CNN model. It is shown in the experiments, that the Top-N cost function offers performance gains in comparison to standard MSE. Further improvement in generalization ability of the network is achieved by using dropout.The second sub-system is a super-resolution deep convolutional network, which performs an enhanced-input associative mapping between input low-resolution and high-resolution images. This network has been trained with low-resolution elevation data and the corresponding high-resolution optical urban photographs. Such a resolution discrepancy between optical aerial/satellite images and elevation data is often the case in real world applications. More specifically, low-resolution elevation data augmented by high-resolution optical aerial photographs are used with the aim of augmenting the resolution of the elevation data. This is a unique super-resolution problem where it was found that many of -the proposed general-image SR propositions do not perform as well. The network aptly named building super resolution CNN (BSRCNN) is trained using patches extracted from the aforementioned data. Results show that in comparison with a classic bicubic upscale of the elevation data the proposed implementation offers important improvement as attested by a modified PSNR and SSIM metric. In comparison, other proposed general-image SR methods performed poorer than a standard bicubic up-scaler.Finally, the relaxation system fuses together all these multisource data sources comprising of pre-processed optical data, elevation data, foliage masks, shadow masks and other pre-processed data in an attempt to assign confidence values to each pixel belonging to a building contour. Confidence is augmented or decremented iteratively until the MSE error fails below a specified threshold or a maximum number of iterations have been executed. The confidence matrix can then be used to extract the true building contours via thresholding

The concept of proximity is an important aspect of human reasoning. Despite the diversity of applications that require proximity measures, the most intuitive notion is that of spatial nearness. The aim of this thesis is to investigate the underpinnings of the notion of nearness, propose suitable formalisations and apply them to the processing of GIS data. More particularly, this work offers a framework for spatial proximity that supports the development of more intuitive tools for users of geographic data processing applications. Many of the existing spatial reasoning formalisms do not account for proximity at all while others stipulate it by using natural language expressions as symbolic values. Some approaches suggest the association of spatial relations with fuzzy membership grades to be calculated for locations in a map using Euclidean distance. However, distance is not the only factor that influences nearness perception. Hence, previous work suggests that nearness should be defined from a more basic notion of influence area. I argue that this approach is flawed, and that nearness should rather be defined from a new, richer notion of impact area that takes both the nature of an object and the surrounding environment into account. A suitable notion of nearness considers the impact areas of both objects whose degree of nearness is assessed. This is opposed to the common approach of only taking one of both objects, seen as a reference to assess the nearness of the other to it, into consideration. Cognitive findings are incorporated to make the framework more relevant to the users of Geographic Information Systems (GIS) with respect to their own spatial cognition. GIS users bring a wealth of knowledge about physical space, particularly geographic space, into the processing of GIS data. This is taken into account by introducing the notion of context. Context represents either an expert in the context field or information from the context field as collated by an expert. In order to evaluate and to show the practical implications of the framework, experiments are conducted on a GIS dataset incorporating expert knowledge from the Touristic Road Travel domain.

La modélisation et la résolution de problèmes sous contraintes constituent un domaine majeur enIA. Par la nature diverse des contraintes, différents formalismes de représentation ont été proposés pour les exprimer de manière simple et compacte tout en garantissant une efficacité des outils de résolutions associés. Les formules propositionnelles, les réseaux de contraintes discrets (RCD) et qualitatives (RCQ) sont des cadres de modélisation répondant à ces critères. Pour les informations temporelles ou spatiales, les RCQ constituent un modèle de choix avec de nombreuses applications comme l’ordonnancement de tâches, la planification temporelle ou spatiale, les systèmes d’informations géographiques. Nos contributions visent à étudier les liens des RCQ vers les RCD et les formules propositionnelles, afin d’adapter les outils issus des divers domaines et de proposer de nouvelles approches. Tout d’abord, nous nous concentrons sur l’aspect structurel des RCQ, en adaptant la méthode de la composition faible dans les différents cadres. Nous exploitons ensuite les propriétés des classes traitables de certains formalismes qualitatifs, afin de définir une transformation vers la logique propositionnelle. En exploitant la transformation vers les RCD, nous proposons une méthode incomplète facilitant la preuve de l’incohérence des RCQ par la relaxation de la propriété de composition faible, puis nous complétons l’approche en exploitant les classes traitables. Enfin, ces études nous conduisent à proposer une nouvelle forme de substituabilité locale, dont les détections statique et dynamique permettent d’obtenir une amélioration algorithmique dans le cadre des RCD
Modelling and solving constraints problems is a major domain in Artificial Intelligence. By the various natures of the constraints, different formalisms were proposed to express them in a simple andcompact way while guaranteeing the effectiveness of the associated solution tools. Propositional formulae, discrete constraint networks (DCNs), and qualitative constraint networks (QCNs) are the well known frameworks that guaranty these requirements. For temporal or space information, QCNs constitute a model of choice with many real world applications such as scheduling, temporal or spatial planning and geographic information systems. Our contributions aim at studying the links between QCNs, DCNs and propositional formulas, in order to adapt the tools developed in these fields and to propose new approaches. First of all, we focus on the structural aspects of QCNs, by transforming weak composition within the various frameworks. In order to define a transformation towards propositional logic we then exploit the properties of tractable classes of some qualitative formalism. Exploiting the transformation towards DCNs, we propose an incomplete method simplifying the proof of the inconsistency for QCNs by relaxing the weak composition property. Then, we propose a complete approach thanks to tractable classes. Finally, these studies lead us to propose a new form of local substitutability, whose static and dynamic detections significantly improve search algorithms for DCNs

Gardner (1985) defines intelligence broadly as the ability to solve problems and create products as well as to find or create new problems. He also suggests that every normal individual has the capacity to develop abilities in seven different areas or types of intelligence. Maker (1992, in press) hypothesizes that gifted individuals competently solve problems of all types, that is, problems ranging from well-defined to ill-defined. In this study of spatial intelligence, the theories of both researchers were tested. The primary purpose of the study was to describe behaviors that could be observed, processes subjects reported using, and characteristics of products subjects created as they solved the series of spatial problems. A secondary purpose of the study was to determine if careful observation of processes subjects employed, combined with the subject's report of processes used and an evaluation of products produced could, indeed, paint a clear picture of the subject's spatial abilities. Six research questions guided the study. The three primary areas of investigation were (a) the similarities and/or differences of behaviors observed, processes reported, and characteristics of products across tasks for individual subjects, (b) the similarities and/or differences of behaviors observed, processes reported, and characteristics of products across subjects for each task, and (c) the similarities and/or differences of behaviors observed, processes reported, and characteristics of products to Gardner's description of spatial intelligence. The researcher delineated eight broad categories of observed behaviors, two broad categories of processes reported, and nine characteristics of finished products. She noted both similarities and differences in behaviors, processes, and products across subjects for tasks and across tasks for subjects. Subjects reported that they used processes similar to those described by Gardner (1985); Gardner does not establish behaviors that can be observed as subjects solve spatial problems, nor does he clearly establish characteristics that might be included in products subjects created. Therefore, the behaviors noted and the characteristics of products created by the subjects in this study allowed the researcher to further define spatial intelligence. (Abstract shortened by UMI.)

Nous nous intéressons à la prise en compte de l'information spatiale d'un point de vue de la représentation et du raisonnement afin de planifier une mission dans le cadre de l'interaction homme-robot. Une première partie de la thèse concerne la représentation et le raisonnement spatial. Nous proposons une ontologie spatiale : SpaceOntology permettant de représenter l'information spatiale qualitative, quantitative et de mener un raisonnement sur celle-ci. Sur la base de cette modélisation spatiale nous avons mis en place un planificateur avec deux modules : un module de raisonnement symbolique supporté par un planificateur de tâches et un module de raisonnement spatial supporté par le planificateur de chemins et SpaceOntology. Enfin, la troisième partie de nos contributions concerne l'extension du langage de planification PDDL à l'information spatiale. Cette extension permet de donner à un problème de planification une sémantique spatiale
We are interested in the integration of spatial information for representation and reasoning to plan a mission in the human-robot interaction case. The first part of our work is for representation and spatial reasoning. We propose a spatial ontology: SpaceOntology qualitative and quantitative representation and reasonning. Based on this spatial modeling we have implemented a planner with two modules: a module for symbolic reasoning supported by a task planner and a module for spatial reasoning supported by the path planning and SpaceOntology. The third part of our contributions is the extension of the PDDL planning language to spatial information. This extension permits to express spatial planning problems

The entire thesis text is included in the research.pdf file; the official abstract appears in the short.pdf file; a non-technical public abstract appears in the public.pdf file. Title from PDF of title page (University of Missouri--Columbia, viewed on January 22, 2010). Thesis advisor: Dr. Marjorie Skubic. Includes bibliographical references.

Le informazioni di tipo geografico caratterizzano più dell'80% dei dati utilizzati nei processi decisionali di ogni grande azienda e la loro pervasività è in costante aumento. La Location Intelligence è un insieme di strumenti, metodologie e processi nati con l'obiettivo di analizzare e comprendere a pieno il patrimonio informativo presente in questi dati geolocalizzati. In questo progetto di tesi si è sviluppato un sistema completo di Location Intelligence in grado di eseguire analisi aggregate dei dati georeferenziati prodotti durante l'operatività quotidiana di una grande azienda multiservizi italiana. L’immediatezza dei report grafici e le comparazioni su serie storiche di diverse sorgenti informative integrate generano un valore aggiunto derivante dalle correlazioni individuabili solo grazie a questa nuova dimensione di analisi. In questo documento si illustrano tutte le fasi caratterizzanti del progetto, dalla raccolta dei requisiti utente fino all’implementazione e al rilascio dell’applicativo, concludendo con una sintesi delle potenzialità di analisi generate da questa specifica applicazione e ai suoi successivi sviluppi.

Le problème étudié est la conception préliminaire d'un bâtiment (allocation spatiale): passage d'un organigramme fonctionnel (modèle topologique) a l'avant-projet sommaire (modèle géométrique). Une maquette est implémentée en genesia 2 (générateur de systèmes experts) faisant le passage de phénix (logiciel CAO de schématique 2d) à PDMS (logiciel CAO 3d). Un bâtiment étant un objet a haut niveau de hiérarchisation, la représentation objet est adoptée. La conception est vue comme un processus de satisfaction de contraintes. La boite (parallélépipède rectangle) est choisie comme objet de base. Ainsi une algèbre (de manhattan) est développée. Le problème de placement sous contraintes d'objets dans l'espace est un problème combinatoire NP-complet lorsqu'on discrétise l'espace. Les contraintes spatiales symboliques deviennent numériques. Un algorithme de résolution de contraintes est écrit avec propagation de contraintes et heuristiques sur les objets

Developing a new spatial model for generating poetic intelligence in response to the already constructed and degrading landscape. The thesis has a simple inquiry: what innovative architectural spatial models can be developed within, or in response to, the townships and degraded land located along the Perth-Kalgoorlie water pipeline in Western Australia, to help instigate a new poetic intelligence when considering architectural making that has a direct relationship with the landscape that it exists within? The thesis begins and ends with a triad relationship between human perception, architectural idea making, and landscape: it begins with observation, engagement and recording and ends with a generative proposition. The thesis articulates how the complexities of a defined site can be recorded and modelled to bind disparate elements into being and therefore model more accurately the wholeness of perception that often drives architectural thinking. Commencing with the lens provided by the Perth-Kalgoorlie water pipeline, the thesis examines a domain in which architecture, landscape, and human action combine to activate our poetic intelligence. The thesis shows that we feel what we think we see, the visible power of man in nature and, the relentlessness of a middle distance that has been constructed around us. Through critical reflection a tremoring occurs, causing powerful new imaginings. The research attempts to visualise the new landscape and show that we help to degrade what we treasure. This moment or realisation can be framed as an aesthetic moment that causes us to think again. The research, formulated as a progressive, heightening of experience, leads the observer from Rambler's Gallery through commonplace territory pointing out observations along the way and then ultimately winds these commonplace observations together to construct a new presentation of the commonplace. The final exhibition announces a new spatial model for generating poetic intelligence in response to the already developed and degrading landscape. The exhibition creates a Speed_Space that posits and tests the essential theme of the research; it is an act of invention that creates new knowledge (the poetic intelligence). The common link between architecture and landscape in this thesis is that both are understood to have been significantly constructed by the human subject and, that this constructed landscape is a finite system and is all that we have. This thesis, through the evidence embodied in SPEED_SPACE offers a mechanism to demonstrate what gaining architectural experience is like; uncoiling into the world, observing, weakening, moving at the limit and then coiling up moments of experience, knowledge and perception to create a force of the imagination that generates new poetic intelligence as a result being in 'that' world. The new spatial model shows architectural experience, in response to the already constructed and degrading landscape, to be more like a self-made constellation acting as a force of imagination rather than a sequence of facts collected together.

There is a recognised need for the continued improvement of both the techniques and technology for spatial decision support in infrastructure site selection. Many authors have noted that current methodologies are inadequate for real-world site selection decisions carried out by heterogeneous groups of decision-makers under uncertainty. Nevertheless despite numerous limitations inherent in current spatial problem solving methods, spatial decision support systems have been proven to increase decision-maker effectiveness when used. However, due to the real or perceived difficulty of using these systems few applications are actually in use to support decision-makers in siting decisions. The most common difficulties encountered involve standardising criterion ratings, and communicating results. This research has focused on the use of Approximate Reasoning to improve the techniques and technology of spatial decision support, and make them easier to use and understand. The algorithm developed in this research (ARAISS) is based on the use of natural language to describe problem variables such as suitability, certainty, risk and consensus. The algorithm uses a method based on type II fuzzy sets to represent problem variables. ARAISS was subsequently incorporated into a new Spatial Decision Support System (InfraPlanner) and validated by use in a real-world site selection problem at Australia's Brisbane Airport. Results indicate that Approximate Reasoning is a promising method for spatial infrastructure planning decisions. Natural language inputs and outputs, combined with an easily understandable multiple decision-maker framework created an environment conducive to information sharing and consensus building among parties. Future research should focus on the use of Genetic Algorithms and other Artificial Intelligence techniques to broaden the scope of existing work.

Nous introduisons les meta-modeles spatiaux pour la representation generique de relations spatiales entre objets manuscrits. Ces modeles peuvent etre appris a partir de donnees et offrent une precision accrue en prenant en compte les formes des objets. Leur definition floue permet la gestion de l’imprecision et offre une description conforme a l’intuition. Un meta-modele a un pouvoir de prediction qui permet de decrire la relation modelisee dans l’image, par rapport a un objet de reference. Cette capacite rend possible la visualisation des modeles et constitue un outil pour segmenter les traces en fonction de leur contexte. Nous proposons une representation des symboles manuscrits structures reposant uniquement la description fine de leur information spatiale par ces modeles. La segmentation des traces en primitives est guidee par la capacite de prediction des modeles. Les resultats confirment l'importance de l'information spatiale et valident la qualite de la modelisation proposee
Spatial meta-models are introduced for a generic description of spatial relations between handwritten objetcs. The models are trainable from data and offer an increased precision thanks to the consideration of objetcs shapes. The fuzzy definition allows to deal with imprecision and provides an intuitive modeling. Meta-models have a predictive ability which enables the description of the modeled relation in the image space, relatively to a reference object. It allows the model visualization and provide a tool for context-driven object segmentation. We proporse a resentation of structured handwritten symbols based only on the description of their spatial information with meta-models. Segmentation of objetcs into primitives is driven by the models prediction ability. Results confirme the importance of spatial information and validate the quality of the proposed models

This thesis presents a method of algorithmic composition in which the music is seen as motion through a multidimensional musical space. An analogy is drawn between physical space and musical space, each direction of the physical space corresponding to a musical parameter. A computer program was developed using the MAX programming environment to simulate the goaldirected motion of a mobile robot through an environment containing obstacles. The potential field method of mobile robot path planning was used. The program maps the location of the robot to musical parameters in the musical space. Based on the instantaneous values of the musical parameters, the program generates melodic material and transmits the resulting MIDI data to a synthesizer. For this research, the program was limited to three spatial dimensions and one obstacle. The program successfully created simple compositions consisting of large musical gestures. A model composition was created. Suggestions were made for further development and more elaborate applications of the method.
School of Music

The topic of the current Master's thesis is Business Intelligence's data presentation using maps. Through integrating BI and geographic information systems a new discipline is emerging - Location Intelligence. The main goal of this thesis is to highlight and analyse reporting possibilities of the BI tools in the framework of maps. The theoretical part of this paper is dedicated to the foundation and principles of geographic information systems and their intersection with BI, where such field as Location Intelligence is being created. In the practical part of the thesis the BI tools IBM Cognos and Oracle BI are compared. The comparison is based on the following criteria: field of implementation, visualization, map external sources and performance. The evaluating criteria are defined in the beginning of the practical part as well as the evaluation method. The methods of analysis and information collecting were used to extract and revise the knowledge from specific electronic and printed sources in Czech or English. Sources for the practical part origin from my technical knowledge of the field of BI, as well as practical experience with implementation of map sources as a feature of Business Intelligence. Statistical methods are used for evaluation of the criteria results. The practical and theoretical value of the thesis lies in creating the lucid comparison of implementation of the map sources into the selected BI tools and options for reporting or visualization of BI data over map. Apart from comparison the framework for implementation of maps into the selected BI tools is established in the above mentioned work.

La finalità del lavoro svolto è volta alla progettazione e successiva implementazione di una Dashboard di Location Intelligence nell'ambito del fashion retail. La Dashboard oltre a fornire una panoramica generale sull’andamento delle vendite (analisi descrittiva), e ad individuare una possibile correlazione delle vendite con dati interni ed esterni (analisi del potenziale), elabora un modello di Machine Learning che combinato con tecniche di Location Intelligence, consente di riconoscere potenziali punti vendita da immettere sul territorio nazionale (analisi predittiva). Il risultato finale, considera il grado di appetibilità del territorio, escludendo le zone che attualmente già offrono soluzioni e che hanno costituito, insieme all’integrazione di Point of Interest, la base per costruire il modello predittivo.

In this dissertation, we consider the clustering problem in data sets with unknown number of clusters having arbitrary shapes, intracluster and intercluster density variations. We introduce a clustering methodology which is composed of three methods that ensures extraction of local density and connectivity properties, data set reduction, and clustering. The first method constructs a unique neighborhood for each data point using the connectivity and density relations among the points based upon the graph theoretical concepts, mainly Gabriel Graphs. Neighborhoods subsequently connected form subclusters (closures) which constitute the skeleton of the clusters. In the second method, the external shape concept in computational geometry is adapted for data set reduction and cluster visualization. This method extracts the external shape of a non-convex n-dimensional data set using Delaunay triangulation. In the third method, we inquire the applicability of Swarm Intelligence to clustering using Ant Colony Optimization (ACO). Ants explore the data set so that the clusters are detected using density break-offs, connectivity and distance information. The proposed ACO-based algorithm uses the outputs of the neighborhood construction (NC) and the external shape formation. In addition, we propose a three-phase clustering algorithm that consists of NC, outlier detection and merging phases. We test the strengths and the weaknesses of the proposed approaches by extensive experimentation with data sets borrowed from literature and generated in a controlled manner. NC is found to be effective for arbitrary shaped clusters, intracluster and intercluster density variations. The external shape formation algorithm achieves significant reductions for convex clusters. The ACO-based and the three-phase clustering algorithms have promising results for the data sets having well-separated clusters.

Real-time strategy (RTS) games constitute one of the largest game genres today and have done so for the past decade. A central feature of real-time strategy games is opponent AI which is suggestively the “last frontier” of game development because the focus of research has primarily been on other components, graphics in particular. This has led to AI research being largely ignored within the commercial game industry but several methods have recently been suggested for improving the strategic ability of AI agents in real-time strategy games.

The aim of this project is to evaluate how a method called qualitative spatial reasoning can improve AI on a tactical level in a selected RTS game. An implementation of an AI agent that uses qualitative spatial reasoning has been obtained and an evaluation of its performance in an RTS game example monitored and analysed.

The study has shown that qualitative spatial reasoning affects AI agent’s behaviour significantly and indicates that it can be used to deduce a rule-base that increases the unpredictability and performance of the agent.

The purpose of the study was to compare 2D and 3D visual presentation styles, both still frame and animation, on subjects' brain activity measured by the amplitude of EEG alpha wave and on their recall to see if alpha power and recall differ significantly by depth and movement of visual presentation style and by spatial intelligence. In addition, the study sought to determine whether there is any significant interaction between spatial intelligence and visual presentation style on alpha power and recall, and to determine whether any relationship exists between alpha power and recall.

The subjects in the present study were one hundred and twenty three undergraduate students at a university in the Midwest. After taking Vandenberg & Kuse's Mental Rotations Test, subjects were divided into low and high spatial intelligence groups, and subjects in each spatial intelligence group were evenly assigned to four different types of visual presentation style (2D still frame, 2D animation, 3D still frame, and 3D animation), receiving an instruction on LASIK eye surgical procedure in its respective visual presentation style. During the one-minute visual instruction, subjects' brain activity was measured and recorded using a wireless EEG headset. Upon completion of the instruction, subjects were given a 10-item multiple-choice test to measure their recall of the material presented during the instruction.

Two 2 (spatial intelligence) x 2 (depth) x 2 (movement) factorial Analysis of Variance (ANOVA) were conducted, one with alpha power as a dependent variable and the other with recall as a dependent variable, to determine whether there is a significant difference in alpha power and recall by spatial intelligence and visual presentation style, as well as whether there is an interaction between these variables that affects alpha power and recall. The Pearson Correlation Coefficient was calculated to examine relationship between alpha power and recall.

The present study found (a) EEG alpha power did not differ by the difference in depth and movement, (b) 2D and animation were found to be more effective on recall, (c) alpha power did not differ by spatial intelligence, (d) recall did not differ by spatial intelligence, (e) there was a significant interaction between spatial intelligence and movement that affected alpha power; still frame resulted in higher alpha power for low spatial learners, and animation resulted in higher alpha power for high spatial learners, (f) there was a significant interaction between spatial intelligence, depth and movement on recall; for low spatial learners, 2D animation resulted in significantly higher recall than both 2D still frame and 3D animation, and for high spatial learners, 3D animation resulted in significantly higher recall than 3D still frame, and both 2D still frame and 2D animation resulted in close to significantly higher recall than 3D still frame, and (g) there was a mildly inverse relationship between alpha power and recall, brought on by a strong inverse relationship in 2D still frame revealing a 'higher alpha power-lower recall connection' for low spatial learners and a 'lower alpha power-higher recall connection' for high spatial learners.

The spatial decision-making process for the purpose of planning, policies, plans, or territorial programs, is intimately tied to the concept of location in the respective geographical space, mainly because many aspects such as natural resources; raw materials, infrastructure, proximity to markets, weather, culture, safety and strategic areas, among others, can be determined, and directly or indirectly related, based on spatial locations. Because of this and assuming that the geographical space constitutes a complex system, it is necessary that the spatial decision-making process intended for achieving aspects of location of goods and services, and/or events for the aforementioned purposes, is carried out in an interdisciplinary way, preferably through participation and consensus from the involved stakeholders involved.In this thesis, an applied research is presented that had as main purpose to provide an archetype that supports the process of spatial decision-making through the spatial consensus by diverse social actors, specially in regard to the location of goods, services and/or events. The archetype, based on the spatial version of the Delphi method modified, has as main characteristic the use of the opinions from a group of experts to solve a problem in the context of the decision-making process and / or in the framework for foresight while also considering the geographical aspect, all of this in order to reach a consensus in regard to the identification of an optimal location in a limited part of a territory.By following the Design Science Research methodology (DSR) and by adopting as auxiliary methods, Documentary Research and Reuse-Based Software Development, an artifact called Geospatial System of Collective Intelligence (SIGIC) was devised through which a collective intelligence behavior could be induced into multidisciplinary groups not experienced in the use of geotechnologies, and for the purpose of locating goods, services, and / or events through the spatial consensus from the participants. After having evaluated the artifact in five different spatial decision situations, two of them in Italy for the location of an air quality monitoring station, and for the zoning of the street prostitution process; two of them in Ecuador to determine the location for the installation of a booth for the screening of HIV and to state the location of booths for the annual initiative for blood collection, and finally, one more in Antarctica, Greenwich Island to determine the location of goods and services to plan the activities of the Scientific Station Pedro Vicente Maldonado.The main results exhibit that even multidisciplinary groups without previous experience in the use of geotechnologies can interact with these tools to study and determine through consensus of their spatial opinions, the potential location of goods, services and / or events in a given territory, thus achieving in this way a process of Collective Spatial Analysis as support for spatial decisions, mainly in aspects related to planning, organization, and use of resources in the territory, and even for the evaluation, preparation, or adoption of policies, plans, and corresponding programs.
El proceso de toma de decisiones espaciales con fines de planificación, políticas, planes o programas territoriales, está estrechamente vinculado con el concepto de la localización en el espacio geográfico respectivo, principalmente porque en base a las ubicaciones espaciales se pueden determinar y relacionar directa e indirectamente, muchos otros aspectos como son los recursos naturales, las materias primas, la infraestructura, la proximidad al mercado, el clima, la cultura, las áreas estratégicas y de seguridad, entre otros. Por esta razón, y asumiendo que el espacio geográfico constituye un sistema complejo, es necesario que la toma de decisiones espaciales destinada a lograr aspectos de localización de bienes servicios y/o eventos para lo fines antes mencionados, se efectúe interdisciplinarmente, preferentemente a través de la participación y consenso de los actores involucrados.En esta tesis se presenta una investigación aplicada que tuvo como propósito aportar un arquetipo para coadyuvar en el proceso de toma de decisiones espaciales por medio del consenso espacial entre diversos actores sociales, primordialmente con respecto a la localización de bienes, servicios y/o eventos. El arquetipo, basado en la versión espacial del Método Delphi modificado, tiene como característica principal el uso de las opiniones de un grupo de expertos para resolver un problema en el contexto de toma de decisiones y/o marco de previsión considerando el aspecto geográfico, con el objetivo de llegar a un consenso respecto a la identificación de una ubicación óptima en una parte limitada del territorio.Siguiendo la metodología de Investigación en Ciencia del Diseño (DSR por su acrónimo en Inglés) y adoptando como métodos auxiliares la Investigación Documental y el Desarrollo de Software Basado en la Reutilización, se ideó el artefacto llamado Sistema Geoespacial de Inteligencia Colectiva (SIGIC), con el cual se logró inducir y coordinar un comportamiento de inteligencia colectiva en grupos multidisciplinares no experimentados en el uso de geotecnologías, con el propósito de localizar bienes, servicios y/o eventos mediante el consenso espacial de los participantes. Este artefacto se evaluó en cinco situaciones diferentes de decisión espacial: dos en Italia, para la ubicación de una estación de monitoreo de calidad del aire y para la zonificación del proceso de prostitución de calle; dos en Ecuador, para determinar el lugar de instalación de un puesto para pruebas de tamizaje de VIH y para establecer las ubicaciones de puestos para la campaña anual de colecta de sangre; y una en la Antártida, Isla de Greenwich, para determinar la ubicación de bienes y servicios con fines de planificación de actividades de la Estación Científica Pedro Vicente Maldonado.Los principales resultados exhiben que grupos multidisciplinares sin práctica en el uso de geotecnologías pueden interactuar con estas herramientas para estudiar y determinar a través del consenso de sus opiniones espaciales, la probable localización de bienes, servicios y/o eventos en un territorio determinado, conformando de esta manera un proceso de Análisis Espacial Colectivo como soporte a las decisiones espaciales principalmente, para aspectos de planificación, organización y empleo de recursos en el territorio o bien para la evaluación, preparación o adopción de políticas, planes y programas correspondientes.
El procés de presa de decisions espacials amb fins de planificació, polítiques, plans o programes territorials, esta íntimament vinculat amb el concepte de la localització en l’espai geogràfic respectiu, principalment perquè en base a las ubicacions espacials es pot determinar i relacionar directament i indirecta, molts altres aspectes com són els recursos naturals, les matèries primeres, la infraestructura, la proximitat al mercat, el clima, la cultura, les àrees estratègiques i de seguretat, entre d´altres. Per aquesta raó, i assolint que el espai geogràfic constitueix un sistema complex, és necessari que la presa de decisions espacials destinada a aconseguir aspectes de localització de béns serveis i/o esdeveniments per els fins anteriorment esmentats, s’efectui interdisciplinariament, de manera preferent mitjançant la participació i consens dels actors involucrats.A aquesta tesi es presenta una investigació aplicada que va tenir com a propòsit aportar un arquetip per coadjuvar en el procés de presa de decisions espacials mitjançant el consens espacial entre diversos actors socials, primordialment respecte de la localització de béns, serveis i/o esdeveniments. L'arquetip, basat en la versió espacial del Mètode Delphi modificat, té com a característica principal l'ús de les opinions d'un grup d'experts per resoldre un problema en el context de presa de decisions i/o marc de previsió considerant l'aspecte geogràfic, amb el objectiu d'arribar a un consens respecte a la identificació d'una ubicació òptima en una part limitada del territori.Tot seguint la metodologia d´Investigació en Ciència del Disseny (DSR per el seu acrònim en anglès) i adoptant com a mètodes auxiliars la Investigació Documental i el Desenvolupament de Software Fonamentat en la Reutilització, es va idear l’artefacte anomenat Sistema Geoespacial d’Intel·ligència Col·lectiva (SIGIC), mitjançant del qual es va aconseguir induir i coordinar un comportament d’intel·ligència col·lectiva en grups multidisciplinaris no experimentats en l’ús de geotecnologíes, amb el propòsit de localitzar béns, serveis i/o esdeveniments mitjançant el consens espacial dels participants. Aquest artefacte es va avaluar en cinc situacions diferents de decisió espacial: dos a Itàlia, per a la ubicació d'una estació de monitorització de qualitat de l'aire i per a la zonificació del procés de prostitució del carrer; dos a Equador, per determinar el lloc d'instal·lació d'un posat per a proves de tamisatge de VIH i per establir les ubicacions de llocs per a la campanya anual de col·lecta de sang; i una a l'Antàrtida, Illa de Greenwich, per determinar la ubicació de béns i serveis amb fins de planificació d'activitats de l'Estació Científica Pedro Vicente Maldonado.Els principals resultats exhibeixen que grups multidisciplinaris sense practica en l’ús de les geotecnologíes poden interactuar amb aquestes eines per estudiar i determinar mitjançant el consens de les seves opinions espacials, la probable localització de béns, serveis i/o esdeveniments en un territori determinat, conformant d´aquesta manera un procés d’Anàlisi Espacial Col·lectiu com a suport a les decisions espacials principalment, per aspectes de planificació, organització i ocupació de recursos en el territori o bé per l´avaluació, preparació o adopció de polítiques, plans i programes corresponents.
Il processo decisionale spaziale ai fini della pianificazione, delle politiche, dei piani o dei programmi territoriali, è strettamente legato al concetto di localizzazione dello spazio geografico di riferimento, principalmente perché in base alle ubicazioni spaziali possono essere determinati e si relazionano direttamente e indirettamente, molti altri aspetti tra i quali le risorse naturali, le materie prime, le infrastrutture, la vicinanza al mercato, il clima, la cultura, le aree strategiche e di sicurezza. Quindi, assumendo che lo spazio geografico costituisce un sistema complesso, è necessario che le decisioni spaziali volte ad ottenere aspetti per la localizzazione di beni e servizi e/o eventi per i fini prima menzionati, siano effettuate in maniera interdisciplinare, preferibilmente attraverso la partecipazione e il consenso tra le parti interessate. In questa tesi viene presentata una ricerca applicata che ha avuto come proposito fornire un archetipo per aiutare nel processo decisionale spaziale per mezzo del consenso speciale tra diversi attori sociali, soprattutto per quanto riguarda la localizzazione dei beni, servizi e/o eventi. L'archetipo, in base alla versione spaziale del Metodo Delphi modificato, ha come caratteristica principale l'uso delle opinioni di un gruppo di esperti per risolvere un problema nel contesto del processo decisionale e/o di un quadro di previsione considerando l'aspetto geografico, con l'obiettivo di giungere ad un consenso riguardo l'identificazione di una posizione ottimale in una parte limitata del territorio. Seguendo la metodologia della Ricerca Scientifica di Progettazione (DSR con il suo acronimo in inglese) e adottando come metodi ausiliari la Ricerca Documentaria e lo Sviluppo del Software Basato sulla Riutilizzazione, è stato messo a punto l’artefatto chiamato Sistema Geo-Spaziale di Intelligenza Collettiva (SIGIC), con il quale è stato possibile indurre e coordinare un comportamento di intelligenza collettiva in gruppi multidisciplinari non sperimentati nell'uso di geo-tecniche, al fine di localizzare beni, servizi e/o eventi attraverso il consenso spaziale dei partecipanti. Questo artefatto è stato valutato in cinque situazioni diverse di decisione spaziale: due in Italia, per l'ubicazione di una stazione di monitoraggio della qualità dell'aria e per la zonizzazione del processo di prostituzione di strada; due in Ecuador, per determinare il luogo di installazione di un centro per i test per l'HIV e per stabilire le ubicazioni di centri per la campagna annuale di donazione del sangue; e una in Antartide, nell'isola Greenwich, per determinare l'ubicazione di beni e servizi al fine di pianificare le attività della Stazione Scientifica Pedro Vicente Maldonado. Dopo aver valutato l'artefatto in cinque differenti situazioni di decisione spaziale, i principali risultati mostrano che i gruppi multidisciplinari senza pratica nell'uso di geo-tecniche possono interagire con questi strumenti per studiare e determinare attraverso il consenso delle loro opinioni spaziali, la probabile localizzazione dei beni, servizi e/o eventi in un determinato territorio, formando così un processo di Analisi Spaziale Collettiva principalmente come supporto alle decisioni spaziali, per gli aspetti di pianificazione, organizzazione e l'uso delle risorse in il territorio o anche per la valutazione, la preparazione o l'adozione di politiche, piani e programmi corrispondenti.

In this study the behaviors observed and the products created by Hispanic and Navajo children, ages four to five, are described. Each child participated in a performance-based assessment of problem solving designed to identify young children with gifts and talents. The assessment process entails careful observation of children's problem solving and resultant products to determine an individual's abilities and interests in each of the intelligences posited by Gardner (1983). All children were videotaped in the classroom while engaged in solving problems on the spatial activity of the assessment. An embedded, multiple case study design (Yin, 1994) was used as the formal research strategy to address the research questions posed. Case study methods and qualitative techniques were used. Thus, a total of eight single case studies were conducted. The primary purpose of the study was to describe in detail the behaviors that could be observed and the characteristics of the products created by young culturally diverse children on a problem solving task involving spatial abilities. A secondary purpose of the study was to determine if careful observation of the spatial problem solving behaviors exhibited by the children, combined with an evaluation of final products constructed, could clarify and extend the understanding of the spatial area of intelligence. The researcher delineated three broad categories of observed behaviors and three broad categories of characteristics of finished products. No sex-related differences were found between young female and male children. However, clear qualitative differences between young Navajo and Hispanic children were found in the spatial problem solving behaviors and in the ways that spatial representations were characterized. Similarly qualitative differences were found in the eye patterns and solution strategies used by young children who were characterized as highly able problem solvers on the spatial task. The behaviors noted and the characteristics of products created by the children in this study allowed the researchers to further define spatial intelligence children across the core capacities of the spatial domain.

Dans cette thèse nous étudions les algorithmes d'apprentissage automatique pour la reconnaissance visuelle. Un accent particulier est mis sur l'apprentissage automatique de représentations, c.à.d. l'apprentissage automatique d'extracteurs de caractéristiques; nous insistons également sur l'apprentissage conjoint de ces dernières avec le modèle de prédiction des problèmes traités, tels que la reconnaissance d'objets, la reconnaissance d'activités humaines, ou la segmentation d'objets. Dans ce contexte, nous proposons plusieurs contributions : Une première contribution concerne les modèles de type bags of words (BoW), où le dictionnaire est classiquement appris de manière non supervisée et de manière autonome. Nous proposons d'apprendre le dictionnaire de manière supervisée, c.à.d. en intégrant les étiquettes de classes issues de la base d'apprentissage. Pour cela, l'extraction de caractéristiques et la prédiction de la classe sont formulées en un seul modèle global de type réseau de neurones (end-to-end training). Deux algorithmes d'apprentissage différents sont proposés pour ce modèle : le premier est basé sur la retro-propagation du gradient de l'erreur, et le second procède par des mises à jour dans le diagramme de Voronoi calculé dans l'espace des caractéristiques. Une deuxième contribution concerne l'intégration d'informations géométriques dans l'apprentissage supervisé et non-supervisé. Elle se place dans le cadre d'applications nécessitant une segmentation d'un objet en un ensemble de régions avec des relations de voisinage définies a priori. Un exemple est la segmentation du corps humain en parties ou la segmentation d'objets spécifiques. Nous proposons une nouvelle approche intégrant les relations spatiales dans l'algorithme d'apprentissage du modèle de prédication. Contrairement aux méthodes existantes, les relations spatiales sont uniquement utilisées lors de la phase d'apprentissage. Les algorithmes de classification restent inchangés, ce qui permet d'obtenir une amélioration du taux de classification sans augmentation de la complexité de calcul lors de la phase de test. Nous proposons trois algorithmes différents intégrant ce principe dans trois modèles : - l'apprentissage du modèle de prédiction des forêts aléatoires, - l'apprentissage du modèle de prédiction des réseaux de neurones (et de la régression logistique), - l'apprentissage faiblement supervisé de caractéristiques visuelles à l'aide de réseaux de neurones convolutionnels
In this thesis, we present our research on visual recognition and machine learning. Two types of visual recognition problems are investigated: action recognition and human body part segmentation problem. Our objective is to combine spatial information such as label configuration in feature space, or spatial layout of labels into an end-to-end framework to improve recognition performance. For human action recognition, we apply the bag-of-words model and reformulate it as a neural network for end-to-end learning. We propose two algorithms to make use of label configuration in feature space to optimize the codebook. One is based on classical error backpropagation. The codewords are adjusted by using gradient descent algorithm. The other is based on cluster reassignments, where the cluster labels are reassigned for all the feature vectors in a Voronoi diagram. As a result, the codebook is learned in a supervised way. We demonstrate the effectiveness of the proposed algorithms on the standard KTH human action dataset. For human body part segmentation, we treat the segmentation problem as classification problem, where a classifier acts on each pixel. Two machine learning frameworks are adopted: randomized decision forests and convolutional neural networks. We integrate a priori information on the spatial part layout in terms of pairs of labels or pairs of pixels into both frameworks in the training procedure to make the classifier more discriminative, but pixelwise classification is still performed in the testing stage. Three algorithms are proposed: (i) Spatial part layout is integrated into randomized decision forest training procedure; (ii) Spatial pre-training is proposed for the feature learning in the ConvNets; (iii) Spatial learning is proposed in the logistical regression (LR) or multilayer perceptron (MLP) for classification

Dans cette thèse, nous présentons des modèles appartenant au domaine de l’intelligence artificielle afin de prédire l’indice magnétique global am à partir des paramètres du vent solaire. Ceci est fait dans l’optique de fournir des modèles opérationnels basés sur les données enregistrées par le satellite ACE situé au point de Lagrange L1. L’indice am ne possède pas à l’heure actuelle de modèles de prédiction. Pour prédire cet indice, nous avons fait appel à des modèles non-linéaires que sont les réseaux de neurones, permettant de modéliser le comportement complexe et non-linéaire de la magnétosphère terrestre. Nous avons dans un premier temps travaillé sur le développement et l’optimisation des modèles de réseaux classiques comme le perceptron multi-couche. Ces modèles ont fait leurs preuves en météorologie spatiale pour prédire aussi bien des indices magnétiques spécifiques à des systèmes de courant comme l’indice Dst, caractéristique du courant annulaire, que des indices globaux comme l’indice Kp. Nous avons en particulier étudié un réseau temporel appelé Time Delay Neural Network (TDNN) et évalué sa capacité à prédire l’indice magnétique am à une heure, uniquement à partir des paramètres du vent solaire. Nous avons analysé la sensibilité des performances des réseaux de neurones en considérant d’une part les données fournies par la base OMNI au niveau de l’onde de choc, et d’autre part des données obtenues par le satellite ACE en L1. Après avoir étudié la capacité de ces réseaux à prédire am, nous avons développé un réseau de neurones encore jamais utilisé en météorologie spatiale, le réseau Long Short Term Mermory ou LSTM. Ce réseau possède une mémoire à court et à long terme, et comme le TDNN, fournit des prédictions de l’indice am uniquement à partir des paramètres du vent solaire. Nous l’avons optimisé afin de modéliser au mieux le comportement de la magnétosphère et avons ainsi obtenu de meilleures performances de prédiction de l'indice am par rapport à celles obtenues avec le TDNN. Nous avons souhaité continuer le développement et l’optimisation du LSTM en travaillant sur l’utilisation de fonctions de couplage en entrée de ce réseau de neurones, et sur le développement de réseaux multisorties pour prédire les indices magnétiques am sectoriels ou aσ, spécifiques à chaque secteur Temps Magnétique Local. Enfin, nous avons développé une nouvelle technique combinant réseau LSTM et processus gaussiens, afin de fournir une prédiction probabiliste jusqu’à six heures des indices magnétiques Dst et am. Cette méthode a été dans un premier temps développée pour l’indice magnétique Dst afin de pouvoir comparer les performances du modèle hybride à des modèles de référence, puis appliquée à l’indice magnétique am
In this thesis, we present models which belongs to the field of artificial intelligence to predict the geomagnetic index am based on solar wind parameters. This is done in terms to provide operational models based on data recorded by the ACE satellite located at the Lagrangian point L1. Currently, there is no model providing predictions of the geomagnetic index am. To predict this index, we have relied on nonlinear models called neural networks, allowing to model the complex and nonlinear dynamic of the Earth’s magnetosphere. First, we have worked on the development and the optimisation of basics neural networks like the multilayer perceptron. These models have proven in space weather to predict geomagnetic index specific to current systems like the Dst index, characteristic of the ring current, as well as the global geomagnetic index Kp. In particular, we have studied a temporal network, called the Time Delay Neural Network (TDNN) and we assessed its ability to predict the geomagnetic index am within one hour, base only on solar wind parameters. We have analysed the sensitivity of neural network performance when considering on one hand data from the OMNI database at the bow shock, and on the other hand data from the ACE satellite at the L1 point. After studying the ability of neural networks to predict the geomagnetic index am, we have developped a neural network which has never been used before in Space Weather, the Long Short Term Memory or LSTM. Like the TDNN, this network provides am prediction based only on solar wind parameters. We have optimised this network to model at best the magnetosphere behaviour and obtained better performance than the one obtained with the TDNN. We continued the development and the optimisation of the LSTM network by using coupling functions as neural network features, and by developing multioutput networks to predict the sectorial am also called aσ, specific to each Magnetical Local Time sector. Finally, we developped a brand new technique combining the LSTM network and gaussian process, to provide probabilistic predictions up to six hours ahead of geomagnetic index Dst and am. This method has been first developped to predict Dst to be able to compare the performance of this model with reference models, and then applied to the geomagnetic index am

Résumé Internet est devenu un moyen de diffusion de l’information géographique par excellence. Il offre de plus en plus de services cartographiques accessibles par des milliers d’internautes à travers le monde. Cependant, la qualité de ces services doit être améliorée, principalement en matière de personnalisation. A cette fin, il est important que la carte générée corresponde autant que possible aux besoins, aux préférences et au contexte de l’utilisateur. Ce but peut être atteint en appliquant les transformations appropriées, en temps réel, aux objets de l’espace à chaque cycle de génération de la carte. L’un des défis majeurs de la génération d’une carte à la volée est la résolution des conflits spatiaux qui apparaissent entre les objets, essentiellement à cause de l’espace réduit des écrans d’affichage. Dans cette thèse, nous proposons une nouvelle approche basée sur la mise en œuvre d’un système multiagent pour la génération à la volée des cartes et la résolution des conflits spatiaux. Cette approche est basée sur l’utilisation de la représentation multiple et la généralisation cartographique. Elle résout les conflits spatiaux et génère les cartes demandées selon une stratégie innovatrice : la génération progressive des cartes par couches d’intérêt. Chaque couche d’intérêt contient tous les objets ayant le même degré d’importance pour l’utilisateur. Ce contenu est déterminé à la volée au début du processus de génération de la carte demandée. Notre approche multiagent génère et transfère cette carte suivant un mode parallèle. En effet, une fois une couche d’intérêt générée, elle est transmise à l’utilisateur. Dans le but de résoudre les conflits spatiaux, et par la même occasion générer la carte demandée, nous affectons un agent logiciel à chaque objet de l’espace. Les agents entrent ensuite en compétition pour l’occupation de l’espace disponible. Cette compétition est basée sur un ensemble de priorités qui correspondent aux différents degrés d’importance des objets pour l’utilisateur. Durant la résolution des conflits, les agents prennent en considération les besoins et les préférences de l’utilisateur afin d’améliorer la personnalisation de la carte. Ils améliorent la lisibilité des objets importants et utilisent des symboles qui pourraient aider l’utilisateur à mieux comprendre l’espace géographique. Le processus de génération de la carte peut être interrompu en tout temps par l’utilisateur lorsque les données déjà transmises répondent à ses besoins. Dans ce cas, son temps d’attente est réduit, étant donné qu’il n’a pas à attendre la génération du reste de la carte. Afin d’illustrer notre approche, nous l’appliquons au contexte de la cartographie sur le web ainsi qu’au contexte de la cartographie mobile. Dans ces deux contextes, nous catégorisons nos données, qui concernent la ville de Québec, en quatre couches d’intérêt contenant les objets explicitement demandés par l’utilisateur, les objets repères, le réseau routier et les objets ordinaires qui n’ont aucune importance particulière pour l’utilisateur. Notre système multiagent vise à résoudre certains problèmes liés à la génération à la volée des cartes web. Ces problèmes sont les suivants : 1. Comment adapter le contenu des cartes, à la volée, aux besoins des utilisateurs ? 2. Comment résoudre les conflits spatiaux de manière à améliorer la lisibilité de la carte tout en prenant en considération les besoins de l’utilisateur ? 3. Comment accélérer la génération et le transfert des données aux utilisateurs ? Les principales contributions de cette thèse sont : 1. La résolution des conflits spatiaux en utilisant les systèmes multiagent, la généralisation cartographique et la représentation multiple. 2. La génération des cartes dans un contexte web et dans un contexte mobile, à la volée, en utilisant les systèmes multiagent, la généralisation cartographique et la représentation multiple. 3. L’adaptation des contenus des cartes, en temps réel, aux besoins de l’utilisateur à la source (durant la première génération de la carte). 4. Une nouvelle modélisation de l’espace géographique basée sur une architecture multi-couches du système multiagent. 5. Une approche de génération progressive des cartes basée sur les couches d’intérêt. 6. La génération et le transfert, en parallèle, des cartes aux utilisateurs, dans les contextes web et mobile.
Abstract Internet is a fast growing medium to get and disseminate geospatial information. It provides more and more web mapping services accessible by thousands of users worldwide. However, the quality of these services needs to be improved, especially in term of personalization. In order to increase map flexibility, it is important that the map corresponds as much as possible to the user’s needs, preferences and context. This may be possible by applying the suitable transformations, in real-time, to spatial objects at each map generation cycle. An underlying challenge of such on-the-fly map generation is to solve spatial conflicts that may appear between objects especially due to lack of space on display screens. In this dissertation, we propose a multiagent-based approach to address the problems of on-the-fly web map generation and spatial conflict resolution. The approach is based upon the use of multiple representation and cartographic generalization. It solves conflicts and generates maps according to our innovative progressive map generation by layers of interest approach. A layer of interest contains objects that have the same importance to the user. This content, which depends on the user’s needs and the map’s context of use, is determined on-the-fly. Our multiagent-based approach generates and transfers data of the required map in parallel. As soon as a given layer of interest is generated, it is transmitted to the user. In order to generate a given map and solve spatial conflicts, we assign a software agent to every spatial object. Then, the agents compete for space occupation. This competition is driven by a set of priorities corresponding to the importance of objects for the user. During processing, agents take into account users’ needs and preferences in order to improve the personalization of the final map. They emphasize important objects by improving their legibility and using symbols in order to help the user to better understand the geographic space. Since the user can stop the map generation process whenever he finds the required information from the amount of data already transferred, his waiting delays are reduced. In order to illustrate our approach, we apply it to the context of tourist web and mobile mapping applications. In these contexts, we propose to categorize data into four layers of interest containing: explicitly required objects, landmark objects, road network and ordinary objects which do not have any specific importance for the user. In this dissertation, our multiagent system aims at solving the following problems related to on-the-fly web mapping applications: 1. How can we adapt the contents of maps to users’ needs on-the-fly? 2. How can we solve spatial conflicts in order to improve the legibility of maps while taking into account users’ needs? 3. How can we speed up data generation and transfer to users? The main contributions of this thesis are: 1. The resolution of spatial conflicts using multiagent systems, cartographic generalization and multiple representation. 2. The generation of web and mobile maps, on-the-fly, using multiagent systems, cartographic generalization and multiple representation. 3. The real-time adaptation of maps’ contents to users’ needs at the source (during the first generation of the map). 4. A new modeling of the geographic space based upon a multi-layers multiagent system architecture. 5. A progressive map generation approach by layers of interest. 6. The generation and transfer of web and mobile maps at the same time to users.

La représentation du temps et de l’espace est une tâche importante dans de nombreux domaines de l’Intelligence Artificielle tels que le traitement du langage naturel, les systèmes d’informations géographiques (GIS), la conception assistée par ordinateur (CAO), la navigation de robots. De nombreux formalismes qualitatifs ont été proposés pour représenter un ensemble d’entités spatiales ou temporelles et leurs relations. La plupart de ces formalismes utilisent des réseaux de contraintes qualitatives (RCQ en abrégé) pour représenter l’ensemble des informations d’un système.Dans certaines applications, en particulier de type multi-agents, plusieurs sources d’informations peuvent chacune fournir un réseau de contraintes qualitatives pour représenter leur connaissance sur l’ensemble des positions relatives d’un ensemble d’objets. La multiplicité des sources d’informations fournissant les RCQ fait que souvent ces RCQ sont conflictuels, et il est alors utile de mettre en oeuvre une méthode de fusion de ces réseaux pour résoudre les conflits. Il s’agit de définir des familles d'opérateurs de fusion prenant en argument un ensemble de RCQ modélisant les informations fournies par les différentes sources et retournant un ensemble cohérent d'informations ou temporelles correspondant aux informations globales déduites à partir de celles des différentes sources.De nombreux opérateurs de fusion ont été définis dans le cadre de la logique propositionnelle. En s’inspirant en partie de ces travaux, nous élaborons des processus de fusion spécifiques aux RCQ et nous en étudions les propriétés logiques
Time and space representation is an important task in many domains such as natural language processing, geographic information systems (GIS), computer vision, robot navigation. Many qualitative approaches have been proposed to represent the spatial or temporal entities and their relations. The majority of these formalisms use qualitative constraints networks (QCNs) to represent information about a system.In some application, e. g. multi-agent systems, spatial or temporal information come from different sources, i. e. each source provides a spatial or temporal QCN representing relative positions between objects. The multiplicity of sources providing spatial or temporal information makes that the underlying QCNs are generally conflicting. Indeed it becomes necessary to solve the conflicts and define a set of consistent spatial or temporal information representing the result of merging.Merging multiple sources information has attracted much attention in the framework of propositional logic. We take an inspiration from these works in order to define some merging process specified to QCNs, and study their logic and computational properties

This thesis investigates the functionality of the units used in connectionist Artificial Intelligence systems. Artificial Neural Networks form the foundation of the research and their units, Artificial Neurons, are first compared with alternative models. This initial work is mainly in the spatial-domain and introduces a new neural model, termed a Taylor Series neuron. This is designed to be flexible enough to assume most mathematical functions. The unit is based on Power Series theory and a specifically implemented Taylor Series neuron is demonstrated. These neurons are of particular usefulness in evolutionary networks as they allow the complexity to increase without adding units. Training is achieved via various traditiona and derived methods based on the Delta Rule, Backpropagation, Genetic Algorithms and associated evolutionary techniques. This new neural unit has been presented as a controllable and more highly functional alternative to previous models. The work on the Taylor Series neuron moved into time-domain behaviour and through the investigation of neural oscillators led to an examination of single-celled intelligence from which the later work developed. Connectionist approaches to Artificial Intelligence are almost always based on Artificial Neural Networks. However, another route towards Parallel Distributed Processing was introduced. This was inspired by the intelligence displayed by single-celled creatures called Protoctists (Protists). A new system based on networks of interacting proteins was introduced. These networks were tested in pattern-recognition and control tasks in the time-domain and proved more flexible than most neuron models. They were trained using a Genetic Algorithm and a derived Backpropagation Algorithm. Termed "Artificial BioChemical Networks" (ABN) they have been presented as an alternative approach to connectionist systems.

The main objective of this thesis is to describe working with spatial data and their use in reports. The author focuses mainly on the MS SQL Server 2008 R2 platform. The first part deals with theoretical foundations. Defines the concept of Location Intelligence and describes its history to the present. It also addresses the benefits of Location Intelligence in various sectors and the relationship with Business Intelligence. Further theoretical part goes on to describe both spatial data at a general level and at the level of MS SQL Server 2008 R2. In addition, it described the use of ESRI Shapefile files and work with these files. From the description of spatial data author goes on to use reporting tools in Microsoft - SQL Server Reporting Services and Report Builder. Possible forms of map visualization and a detailed description of the settings are at the end of the theoretical part. The second part is focused on solving practical problems. Tools that are described in the theoretical part, are used to create report examples that the author at the beginning of this section set out to create.

The specialization of the final master thesis is theory and a practical view of Location Intelligence. The author is writing about specific parts of development of the application which is focused on analysis of consumer's behaviour for managers of retail businesses focused on customer's satisfaction. The application is a new way of researching consumer's behaviour. Spatial data is connected with the subject of research -- analysis of the movement of consumers in retail stores. The application is a practical tool for better targeting for managers or marketing specialists. The app was developed by four researcher Martin Matějka, Tomáš Rosenberger, Vilém Kalus (three diploma researchers) and Ondřej Funiok. The Main goal of the final thesis was to describe the application from the business point of view and to design layout of the user interface. The business model helped us to specify the application from the business point of view. We reached the main goal with research articles of Location Intelligence. The thesis contains theory basis of informatics and geo-informatics field, Business Intelligence, Location Intelligence and selected business models. The practical part of the thesis is composed of integrating application to the business model, functionality description. From functionality description we designed the user interface. The practical part closes with the use case of application in Vysoká škola ekonomická v Praze. Benefits of the thesis are in development of real, new application for future commercial use. Contribution to the theory of Location Intelligence and analytical solution of development of user interface.

The Spatial Reuse (SR) operation is gaining momentum in the latest IEEE 802.11 family of standards due to the overwhelming requirements posed by next-generation wireless networks. In particular, the rising traffic requirements and the number of concurrent devices compromise the efficiency of increasingly crowded Wireless Local Area Networks (WLANs) and throw into question their decentralized nature. The SR operation, initially introduced by the IEEE~802.11ax-2021 amendment and further studied in IEEE 802.11be-2024, aims to increase the number of concurrent transmissions in an Overlapping Basic Service Set (OBSS) using sensitivity adjustment and transmit power control, thus improving spectral efficiency. Our analysis of the SR operation shows outstanding potential in improving the number of concurrent transmissions in crowded deployments, which contributed to enabling low-latency next-generation applications. However, the potential gains of SR are currently limited by the rigidity of the mechanism introduced for the 11ax, and the lack of coordination among BSSs implementing it. The SR operation is evolving towards coordinated schemes where different BSSs cooperate. Nevertheless, coordination entails communication and synchronization overhead, which impact on the performance of WLANs remains unknown. Moreover, the coordinated approach is incompatible with devices using previous IEEE 802.11 versions, potentially leading to degrading the performance of legacy networks. For those reasons, in this thesis, we start assessing the viability of decentralized SR, and thoroughly examine the main impediments and shortcomings that may result from it. We aim to shed light on the future shape of WLANs concerning SR optimization and whether their decentralized nature should be kept, or it is preferable to evolve towards coordinated and centralized deployments. To address the SR problem in a decentralized manner, we focus on Artificial Intelligence (AI) and propose using a class of sequential learning-based methods, referred to as Multi-Armed Bandits (MABs). The MAB framework suits the SR problem because it addresses the uncertainty caused by the concurrent operation of multiple devices (i.e., multi-player setting) and the lack of information in decentralized deployments. MABs can potentially overcome the complexity of the spatial interactions that result from devices modifying their sensitivity and transmit power. In this regard, our results indicate significant performance gains (up to 100\% throughput improvement) in highly dense WLAN deployments. Nevertheless, the multi-agent setting raises several concerns that may compromise network devices' performance (definition of joint goals, time-horizon convergence, scalability aspects, or non-stationarity). Besides, our analysis of multi-agent SR encompasses an in-depth study of infrastructure aspects for next-generation AI-enabled networking.
L'operació de reutilització espacial (SR) està guanyant impuls per a la darrera família d'estàndards IEEE 802.11 a causa dels aclaparadors requisits que presenten les xarxes sense fils de nova generació. En particular, la creixent necessitat de tràfic i el nombre de dispositius concurrents comprometen l'eficiència de les xarxes d'àrea local sense fils (WLANs) cada cop més concorregudes i posen en dubte la seva naturalesa descentralitzada. L'operació SR, inicialment introduïda per l'estàndard IEEE 802.11ax-2021 i estudiada posteriorment a IEEE 802.11be-2024, pretén augmentar el nombre de transmissions concurrents en un conjunt bàsic de serveis superposats (OBSS) mitjançant l'ajustament de la sensibilitat i el control de potència de transmissió, millorant així l'eficiència espectral. El nostre estudi sobre el funcionament de SR mostra un potencial destacat per millorar el nombre de transmissions simultànies en desplegaments multitudinaris, contribuint així al desenvolupament d'aplicacions de nova generació de baixa latència. Tot i això, els beneficis potencials de SR són actualment limitats per la rigidesa del mecanisme introduït per a l'11ax, i la manca de coordinació entre els BSS que ho implementen. L'operació SR evoluciona cap a esquemes coordinats on cooperen diferents BSS. En canvi, la coordinació comporta una sobrecàrrega de comunicació i sincronització, el qual té un impacte en el rendiment de les WLAN. D'altra banda, l'esquema coordinat és incompatible amb els dispositius que utilitzen versions anteriors IEEE 802.11, la qual cosa podria deteriorar el rendiment de les xarxes ja existents. Per aquests motius, en aquesta tesi s'avalua la viabilitat de mecanismes descentralitzats per a SR i s'analitzen minuciosament els principals impediments i mancances que se'n poden derivar. El nostre objectiu és donar llum a la futura forma de les WLAN pel que fa a l?optimització de SR i si s'ha de mantenir el seu caràcter descentralitzat, o bé és preferible evolucionar cap a desplegaments coordinats i centralitzats. Per abordar SR de forma descentralitzada, ens centrem en la Intel·ligència Artificial (AI) i ens proposem utilitzar una classe de mètodes seqüencials basats en l'aprenentatge, anomenats Multi-Armed Bandits (MAB). L'esquema MAB s'adapta al problema descentralitzat de SR perquè aborda la incertesa causada pel funcionament simultani de diversos dispositius (és a dir, un entorn multi-jugador) i la falta d'informació que se'n deriva. Els MAB poden fer front a la complexitat darrera les interaccions espacials entre dispositius que resulten de modificar la seva sensibilitat i potència de transmissió. En aquest sentit, els nostres resultats indiquen guanys importants de rendiment (fins al 100 \%) en desplegaments altament densos. Tot i això, l'aplicació d'aprenentatge automàtic amb múltiples agents planteja diversos problemes que poden comprometre el rendiment dels dispositius d'una xarxa (definició d'objectius conjunts, horitzó de convergència, aspectes d'escalabilitat o manca d'estacionarietat). A més, el nostre estudi d'aprenentatge multi-agent per a SR multi-agent inclou aspectes d'infraestructura per a xarxes de nova generació que integrin AI de manera intrínseca.

Thesis(M.A.)--Case Western Reserve University, 2010
Title from PDF (viewed on 2010-01-28) Department of Psychology Includes abstract Includes bibliographical references and appendices Available online via the OhioLINK ETD Center

A thesis submitted in partial fulfillment of the requirements for the degree of Doctor in Information Management, specialization in Geographic Information Systems
We are living in (smart) cities that hold social-oriented promises but currently, most of these cities disregard the humans. Although some alternatives are appearing such as smart citizen-centric approaches, there is a lack of how promoting truly appealing perspectives toward a common good or better social synergies. Thereby, smart cities, with their associated Information and Communication Technology tools, are offering new possibilities, but, unfortunately, citizens are not fully exploiting the opportunities to empower themselves because, among other reasons, they are not aware of their common spatialities. Currently, we are not able to operationalize the spatial humanurban interactions regarding citizens’ cognitions, feelings and behaviors towards city places (i.e., sense of place) and meaningful geographic human relationships (i.e., social capital). Both concepts are significant as resources for an alternative landscape based on human perception and organization of social interactions fostered through the geographic place(s). In this research, we highlight the need to understand and operationalize social concepts spatial dimension for a better understanding of a smart citizen-centric approach which is mainly dependent on our capability to understand platial urban dynamics. We conceptualized a (spatial) conceptual framework for sense of place and social capital at the individual level to study their spatial relationship in the urban context. We developed a web map-based survey based on the literature to spatialize, characterize and measure sense of place, social capital and civic engagement. Using the spatial data collected, we validated our framework and demonstrated the importance to encompass the spatial dimension of social concepts (i.e., sense of place and social capital) as pivotal aspect (1) to understand the platial urban dynamics; (2) to provide useful social-spatial data to city processes (e.g. civic engagement); and (3) to reveal the potential to include them in social theory and structural equation models. Furthermore, we highlighted the crucial role of Geographic Information Science (GISc) techniques to gather the spatial dimension of those social concepts. Although in this research we focus on the spatial relationship between sense of place and social capital on civic engagement, the possibilities to relate our framework and methodology to other city based-notions can bring to light new platial urban dynamics. This research wants to open up the agenda for further research into exploratory place-based geography studies and, simultaneously, sets up a common social ground to build other socially-oriented conceptualizations or applications on top of it.

Dans le cadre du raisonnement spatial qualitatif, nous utilisons des formalismes axiomatiques définissant des relations méréotopologiques ayant comme primitive spatiale la région étendue plutôt que le point de la géométrie euclidienne. En ajoutant une primitive morphologique, la sphère, et la géométrie des corps de Tarski, nous exprimons des notions d'angle, d'orientation et de distance qualitative, définissant ainsi une méréogéométrie. Dans ce cadre, nous abordons le problème de la formalisation du mouvement ainsi que celui de la forme d'un objet tridimensionnel. Puis nous utilisons la transformée d'axe médian pour construire un graphe de structure de la forme basé sur des sphères caractéristiques que l'on peut aussi définir grâce à la méréogéométrie. Enfin, des outils de théorie des graphes, calcul du groupe d'automorphisme et test d'isomorphisme permettent d'analyser les graphes de structure, tester la similarité de formes ou archiver ces formes dans une base de données.

A geração de expressão de referência (GER) é um dos principais componentes de aplicações de geração de língua natural a partir de dados não linguísticos. Trabalhos existentes nesta área tendem a se concentrar em domínios bidimensionais e tridimensionais simples. Domínios mais realistas, entretanto, ainda são pouco explorados pela literatura, possivelmente pela dificuldade em produzir modelos computacionais com o grau de complexidade exigido. O recente surgimento de aplicações baseadas em ambientes interativos tridimensionais, todavia, oferece uma ampla gama de oportunidades de pesquisa em GER. Nesse trabalho apresentamos um algoritmo de GER para seleção de atributos espaciais em ambientes interativos tridimensionais do tipo GIVE. A solução proposta é uma extensão de um algoritmo de GER tradicional que utiliza relações espaciais combinado com conhecimento linguístico extraído de corpora e modelos computacionais de referência espacial. Essa solução foi avaliada de forma intrínseca no domínio de instruções em mundos virtuais.
Referring expressions generation (REG) is one of the main components in the generation of Natural Language from non-linguistic data. Existing work in the eld tends to focus on simple two- or three-dimensional domains. More realistic domains, however, are still little investigated, possibly due to the diculties in producing computational models with the required degree of complexity. The recent rise of applications based on three-dimensional interactive environments, however, oers a wide range of research opportunities in REG. In this work we present a REG algorithm for selecting spatial attributes in three-dimensional interactive environments GIVE. The proposed solution is an extension of a traditional REG algorithm that makes use of spatial relations combined with linguistic knowledge extracted from corpora and computational models of spatial reference. The proposal was evaluated intrinsically in the domain of instructions generation in virtual worlds.

Thèse (Ph. D.)--Université Laval, 2009.
Texte en anglais avec résumés en anglais et en français. Bibliogr.: f. 185-197. Publié aussi en version électronique dans la Collection Mémoires et thèses électroniques.

Cette thèse présente l'étude et la réalisation d'un Système Expert aidant à la conception d'une loi de pilotage pour un lanceur spatial. Le pilotage est la manœuvre des mouvements de la fusée autour de son centre de gravité. Concevoir une loi comporte des étapes de modélisation, réduction, discrétisation, détermination du contrôleur et de l'estimateur, de validation. Les procédures algorithmiques sont complétées par des connaissances d'origine expérimentale. Aucune solution uniquement algorithmique n'existe dans ces problèmes multicritères analogues à une conception assistée en ingénierie. Les arguments complémentaires favorables à l'utilisation de la méthodologie et des outils des Systèmes Experts, font l'objet de la première partie de la thèse. La deuxième partie analyse les divers types de connaissances du domaine ct du problème : connaissances numériques, théorèmes, connaissances empiriques. Elles sont regroupées en "Sources de Connaissances" indépendantes. Cette structure est adaptée à l'existence d'expertises autonomes relatives à la résolution de sous-problèmes distincts. La troisième partie présente l'architecture du système informatique réalisé et justifie les choix effectués. En particulier, la partie dynamique du système qui est gérée par une structure de type "tableau noir" ("blackboard") : espace de travail commun par lequel les sources de connaissances communiquent entre elles. Le recensement des objets manipulés par les ingénieurs, la présentation de l'arbre de résolution, la liste des algorithmes et la description d'une session commentée sont regroupés dans les annexes.

Face aux multiples applications industrielles, en electronique, mecanique, architecture, urbanisme, l'automatisation du processus d'amenagement spatial suscite un interet considerable. Dans cette these, nous abordons les problemes de placement dans le plan d'objets de forme quelconque lies par des relations locales et de nature diverses. Nous proposons d'associer des techniques de satisfaction de contraintes a une modelisation des espaces de configurations permettant de representer les positions et orientations candidates des objets a placer. En outre, notre approche prend en compte les aspects evolutif, conflictuel et combinatoire du probleme: des algorithmes de consistance dynamique autorisent la resolution des problemes dont les specifications sont mises a jour au cours du processus de resolution par ajout et/ou retrait de contraintes; une methode generique de traitement d'echecs effectue des substitutions des contraintes conflictuelles en minimisant une mesure de frustration: une technique de resolution a precision variable permet une recherche plus efficace en faisant varier la precision geometrique de la solution. Ces principes sont mis en uvre dans le systeme atlas, experimente sur une application de conception d'equipements electroniques hyperfrequences

Les plateformes de crowdsourcing spatial (PCS) sont des systèmes qui permettent à des personnes, appelées commanditaires, de publier des tâches spatiales afin de trouver la main-d’œuvre pour les exécuter. Ces tâches spatiales exigent que leurs exécutants soient à un endroit donné, souvent dans une fenêtre de temps donnée, pour être accomplies. Quelques exemples de PCS sont Uber et TaskRabbit. Les PCS suscitent beaucoup d’intérêt dans la recherche, mais des pistes de recherche sont encore à explorer.Doan et al. [2011] a soutenu que l’objectif réside maintenant dans “la construction de plateformes générales de crowdsourcing qui peuvent être utilisées pour développer rapidement ces systèmes”. Depuis, peu de travaux ont porté sur la conception technique des PCS. En outre, il existe un écart entre ce qui est mis en oeuvre par les PCS de l’industrie et les propositions que l’on trouve dans la littérature scientifique. Nous proposons GENIUS-C, une architecture générique pour les PCS. Nous fournissons une implémentation de référence (IR) pour GENIUS-C, fonctionnant comme un cadre pour le développement de PCS. GENIUS-C et son IR sont destinés à combler les écarts entre le monde académique et industriel, et faciliter la compréhension et le développement rapide de PCS.Nous étudions également l’important problème de l’appariement des exécutants et des tâches d’un PCS. Comment peut-on trouver une ou plusieurs tâches adaptées à un exécutant (et vice versa)? Certains utilisent des techniques de système de recommandation, d’autres des approches d’optimisation. La plupart d’entre eux ne tiennent pas compte des dimensions spatio-temporelles des tâches et des exécutants. Ceux qui en tiennent compte ignorent les préférences des exécutants, des commanditaires ou du système lui-même. Dans ce contexte, nous identifions et modélisons le problème réel et récurrent suivant: une fois que l’exécutant est prêt à accomplir des tâches, quelle est la meilleure séquence de tâches à suivre en respectant ses contraintes spatio-temporelles? Comment cette séquence peut-elle être obtenue en tenant compte des préférences de l’exécutant, des commanditaires, du système lui-même ou d’une combinaison de ceux-ci? Nous nommons cette situation le Problème de la Recommandation de Trajectoire, auquel nous proposons une solution optimale, et étudions des heuristiques d’approximation pour le résoudre
Spatial Crowdsourcing Platforms (SCP) are systems that allow people, called requesters, to publish spatial tasks in order to find suitable workforce to perform it. These spatial tasks require workers to be at a given location, usually within a given time window, to be accomplished. Some examples of SCPs are: Uber, BlaBlaCar and TaskRabbit. SCPs are source of much interest for academy, however several research opportunities remain.Doan et al. [2011] argued that the race is now on “toward building general crowdsourcing platforms that can be used to develop such systems quickly". Since then, little has been done to investigate the technical design of SCPs precisely. Also, there is a gap between what is done in commercial platform and in scientific literature. We propose GENIUS-C, a generic architecture for SPCs. We provide a reference implementation (RI) for GENIUS-C, that works as a framework for the development of SCPs. GENIUS-C and its RI are meant to fill the gap between the academic and industry world, and facilitate the understanding and the quick development of new SCPs.We also study the important problem of matching workers and tasks. How can we find one or more tasks suitable for a worker (and vice versa)? Some tackle this issue using recommender system techniques, others optimization approaches. Most of them do not take into account the spatiotemporal dimensions of tasks and workers. Others take it into account, but to ignore the preferences of either workers, requesters or the system itself. In this context, we identify and model the following common real-life problem: once a worker is willing to spend sometime accomplishing tasks, what is the best sequence of tasks to be followed respecting their spatiotemporal constraints? How can this sequence be obtained taking into account the preferences of the worker, the requesters, the system itself, or a combination of them? We name this situation the Trajectory Recommendation Problem (TRP), propose a feasible exact solution and study approximation heuristics for it

Le raisonnement spatial et temporel qualitatif est un domaine principal d’études de l’intelligence artificielle et, en particulier, du domaine de la représentation des connaissances, qui traite des concepts cognitifs fondamentaux de l’espace et du temps de manière abstraite. Dans notre thèse, nous nous focalisons sur les formalismes du domaine du raisonnement spatial et temporel qualitatif représentant les informations par des contraintes et apportons des contributions sur plusieurs aspects. En particulier, étant donnée des bases de connaissances d’informations qualitatives sur l’espace ou le temps, nous définissons des nouvelles conditions de consistance locale et des techniques associées afin de résoudre efficacement les problèmes fondamentaux se posant. Nous traitons notamment du problème de la satisfiabilité qui est le problème de décider s’il existe une interprétation quantitative de toutes les entités satisfaisant l’ensemble des contraintes qualitatives. Nous considérons également le problème de l’étiquetage minimal qui consiste à déterminer pour toutes les contraintes qualitatives les relations de base participant à au moins une solution ainsi que le problème de redondance consistant à déterminer les contraintes qualitatives non redondantes. En outre, nous enrichissons le domaine des formalismes spatio-temporels par des contributions concernant une logique spatio-temporelle combinant la logique temporelle propositionnelle (PTL) avec un langage de contraintes qualitatives spatiales et une étude de la problématique consistant à gérer une séquence temporelle de configurations spatiales qualitatives devant satisfaire des contraintes de transition
Qualitative Spatial and Temporal Reasoning is a major field of study in Artificial Intelligence and, particularly, in Knowledge Representation, which deals with the fundamental cognitive concepts of space and time in an abstract manner. In our thesis, we focus on qualitative constraint-based spatial and temporal formalisms and make contributions to several aspects. In particular, given a knowledge base of qualitative spatial or temporal information, we define novel local consistency conditions and related techniques to efficiently solve the fundamental reasoning problems that are associated with such knowledge bases. These reasoning problems consist of the satisfiability problem, which is the problem of deciding whether there exists a quantitative interpretation of all the entities of a knowledge base such that all of its qualitative relations are satisfied by that interpretation, the minimal labeling problem, which is the problem of determining all the atoms for each of the qualitative relations of a knowledge base that participate in at least one of its solutions, and the redundancy problem, which is the problem of obtaining all the non-redundant qualitative relations of a knowledge base. Further, we enrich the field of spatio-temporal formalisms that combine space and time in an interrelated manner by making contributions with respect to a qualitative spatio-temporal logic that results by combining the propositional temporal logic (PTL) with a qualitative spatial constraint language, and by investigating the task of ordering a temporal sequence of qualitative spatial configurations to meet certain transition constraints

L'essor des interfaces homme-machine permettant la saisie d'informations à l'aide d'un stylo électronique est accompagné par le développement de méthodes automatiques pour interpréter des données de plus en plus riches et complexes : texte manuscrit, mais aussi expressions mathématiques, schémas, prise de notes libre... Pour interpréter efficacement ces documents manuscrits, il est nécessaire de considérer conjointement les formes des objets qui les constituent et leur positionnement spatial. Nos recherches se concentrent sur la modélisation du positionnement spatial entre des objets manuscrits, en partant du constat qu'il n'est pas exploité dans toute sa richesse par les méthodes actuelles. Nous introduisons le concept de méta-modèle spatial, une modélisation générique pour décrire des relations spatiales entre des objets de nature, complexité et formes variables. Ces modèles, qui peuvent être appris à partir de données, offrent une richesse et une précision inédite car ils autorisent la conduite d'un raisonnement spatial directement dans l'espace image. L'appui sur le cadre de la théorie des sous-ensembles flous et de la morphologie mathématique permet la gestion de l'imprécision et offre une description des relations spatiales conforme à l'intuition. Un méta-modèle est doté d'un pouvoir de prédiction qui permet de décrire la relation spatiale modélisée au sein de l'image, par rapport à un objet de référence. Cette capacité rend possible la visualisation des modèles et fournit un outil pour segmenter les tracés en fonction de leur contexte. En exploitant ces modèles, nous proposons une représentation pour des objets manuscrits à la structure complexe. Cette représentation repose uniquement sur la modélisation de leurs informations spatiales, afin de démontrer l'importance de ces informations pour l'interprétation d'objets manuscrits structurés. La segmentation des tracés en primitives structurelles est guidée par les modèles de positionnement, via leur capacité de prédiction. Les résultats expérimentaux, portant sur des objets de complexité et de natures diverses (caractères chinois, gestes d'édition, symboles mathématiques, lettres), confirment la bonne qualité de description du positionnement offerte par les méta-modèles. Les tests de reconnaissance de symboles par l'exploitation de leur information spatiale attestent d'une part de l'importance de cette information et valident d'autre part la capacité des méta-modèles à la représenter avec une grande précision. Ces résultats témoignent donc de la richesse de l'information spatiale et du potentiel des méta-modèles spatiaux pour l'amélioration des techniques de traitement du document manuscrit.

Cette thèse traite de la problématique de l'unification de différents flux d'informations modales qui peuvent provenir des senseurs d'un agent. Cette unification, inspirée des expériences psychologiques comme l'effet ventriloque, s'appuie sur la détection de corrélations, définies comme des motifs spatiaux qui apparaissent régulièrement dans les flux d'entrée. L'apprentissage de l'espace des corrélations du flux d'entrée échantillonne cet espace et généralise les échantillons appris. Cette thèse propose des principes fonctionnels pour le traitement multimodal de l'information qui ont aboutit à l'architecture connexionniste, générique, modulaire et cortico-inspirée SOMMA (Self-Organizing Maps for Multimodal Association). Dans ce modèle, le traitement de chaque modalité s'effectue au sein d'une carte corticale. L'unification multimodale de l'information est obtenue par la mise en relation réciproque de ces cartes. L'échantillonnage et la généralisation des corrélations reposent sur une auto-organisation contrainte des cartes. Ce modèle est caractérisé par un apprentissage progressif de ces propriétés fonctionnelles: les propriétés monomodales amorcent l'émergence des propriétés multimodales et, dans le même temps, l'apprentissage de certaines corrélations par chaque carte est un préalable à l'auto-organisation de ces cartes. Par ailleurs, l'utilisation d'une architecture connexionniste et d'un apprentissage continu et non supervisé fournit au modèle des propriétés de robustesse et d'adaptabilité qui sont généralement absentes des approches informatiques classiques.