Dissertations / Theses on the topic 'Traces clustering'

Le déboggage des applications de streaming qui s'exécutent sur les systèmes embarqués multimédia est l'un des domaines les plus exigeants dans le développement de logiciel embarqué. Les nouvelles générations de materiel embarqué introduisent de nouvelles systèmes sur une puce, qui fait que les développeurs du logiciel doivent adapter leurs logiciels aux nouvelles platformes. Le logiciel embarqué doit non seulement fournir des résultats corrects mais aussi le faire en temps réel afin de respecter les propriétés de qualité de service (Quality-of-Service, QoS) du système. Lorsque les propriétés QoS ne sont pas respectées, des bugs temporels font leur apparition. Ces bugs se manifestent comme, par exemple, des glitches dans le flux vidéo ou des craquements dans le flux audio. Le déboggage temporel est en général difficile à effectuer car les bugs temporels n'ont pas souvent de rapport avec l'exactitude fonctionnelle du code des applications, ce qui rend les outils de débogage traditionels, comme GDB, peu utiles. Le non-respect des propriétés QoS peut provenir des interactions entre les applications, ou entre les applications et les processus systèmes. Par conséquent, le contexte d'exécution entier doit être pris en compte pour le déboggage temporel. Les avancements récents en collecte des traces d'exécution permettent aux développeurs de recueillir des traces et de les analyser après la fin d'exécution pour comprendre quelle activité système est responsable des bugs temporels. Cependant, les traces d'exécution ont une taille conséquente, ce qui demande aux devéloppeurs des connaissainces en analyse de données qu'ils n’ont souvent pas.Dans cette thèse, nous proposons SATM - une approche novatrice pour le déboggage temporel des applications de streaming. SATM repose sur la prémisse que les applications sont conçues avec le modèle dataflow, i.e. peuvent être représentées comme un graphe orienté où les données sont transmises entre des unités de calcul (fontions, modules, etc.) appelées "acteurs". Les acteurs doivent être exécutés de manière périodique afin de respecter les propriétés QoS représentées par les contraintes de temps-réél. Nous montrons qu'un acteur qui ne respecte pas de façon répétée sa période pendant l'exécution de l'application cause la violation des contraintes temps-reel de l'application. En pratique, SATM est un workflow d'analyse de données venant des traces d'exécution qui combine des mesures statistiques avec des algorithmes de fouille de données. SATM fournit une méthode automatique du débogage temporel des applications de streaming. Notre approche prend en entrée une trace d'exécution d'une application ayant une QoS basse ainsi qu'une liste de ses acteurs, et tout d'abord détecte des invocations des acteurs dans la trace. SATM découvre ensuite les périodes des acteurs ainsi que les séctions de la trace où la période n'a pas été respectée. Enfin, ces séctions sont analysées afin d'extraire des motifs de l'activité système qui différencient ces sections des autres séctions de la trace. De tels motifs peuvent donner des indices sur l'origine du problème temporel dans le systeme et sont rendus au devéloppeur. Plus précisément, nous représentons ces motifs comme des séquences contrastes minimales et nous étudions des différentes solutions pour fouiller ce type de motifs à partir des traces d'exécution.Enfin, nous montrons la capacité de SATM de détecter une perturbation temporelle injectée artificiellement dans un framework multimedia GStreamer, ainsi que des bugs temporels dans deux cas d'utilisation des applications de streaming industrielles provenant de la société STMicroelectronics. Nous fournissons également une analyse détaillée des algorithmes de fouille de motifs séquentiels appliqués sur les données venant des traces d'exécution, et nous expliquons pour quelle est la raison les algorithmes de pointe n'arrivent pas à fouiller les motifs séquentiels à partir des traces d'exécution de façon efficace
Debugging streaming applications run on multimedia embedded systems found in modern consumer electronics (e.g. in set-top boxes, smartphones, etc) is one of the most challenging areas of embedded software development. With each generation of hardware, more powerful and complex Systems-on-Chip (SoC) are released, and developers constantly strive to adapt their applications to these new platforms. Embedded software must not only return correct results but also deliver these results on time in order to respect the Quality-of-Service (QoS) properties of the entire system. The non-respect of QoS properties lead to the appearance of temporal bugs which manifest themselves in multimedia embedded systems as, for example, glitches in the video or cracks in the sound. Temporal debugging proves to be tricky as temporal bugs are not related to the functional correctness of the code, thus making traditional GDB-like debuggers essentially useless. Violations of QoS properties can stem from complex interactions between a particular application and the system or other applications; the complete execution context must be, therefore, taken into account in order to perform temporal debugging. Recent advances in tracing technology allow software developers to capture a trace of the system's execution and to analyze it afterwards to understand which particular system activity is responsible for the violations of QoS properties. However, such traces have a large volume, and understanding them requires data analysis skills that are currently out of the scope of the developers' education.In this thesis, we propose SATM (Streaming Application Trace Miner) - a novel temporal debugging approach for embedded streaming applications. SATM is based on the premise that such applications are designed under the dataflow model of computation, i.e. as a directed graph where data flows between computational units called actors. In such setting, actors must be scheduled in a periodic way in order to meet QoS properties expressed as real-time constraints, e.g. displaying 30 video frames per second. We show that an actor which does not eventually respect its period at runtime causes the violation of the application’s real-time constraints. In practice, SATM is a data analysis workflow combining statistical measures and data mining algorithms. It provides an automatic solution to the problem of temporal debugging of streaming applications. Given an execution trace of a streaming application exhibiting low QoS as well as a list of its actors, SATM firstly determines exact actors’ invocations found in the trace. It then discovers the actors’ periods, as well as parts of the trace in which the periods are not respected. Those parts are further analyzed to extract patterns of system activity that differentiate them from other parts of the trace. Such patterns can give strong hints on the origin of the problem and are returned to the developer. More specifically, we represent those patterns as minimal contrast sequences and investigate various solutions to mine such sequences from execution trace data.Finally, we demonstrate SATM’s ability to detect both an artificial perturbation injected in an open source multimedia framework, as well as temporal bugs from two industrial use cases coming from STMicroelectronics. We also provide an extensive analysis of sequential pattern mining algorithms applied on execution trace data and explain why state-of-the-art algorithms fail to efficiently mine sequential patterns from real-world traces

Notre travail de recherche explore de manière comparée le développement du neuromarketing aux Etats-Unis et en France. Nous commençons par analyser la littérature sur le neuromarketing. Nous utilisons comme cadre théorique et méthodologique l’Actor Network Theory (ANT) ou Théorie de l’Acteur-Réseau (dans le sillage des travaux de Bruno Latour et Michel Callon). Nous montrons ainsi comment des actants « humains et non-humains »: acteurs-réseaux, traces (publications) et controverses forment les piliers d’une nouvelle discipline telle que le neuromarketing. Notre approche hybride « qualitative-quantitative », nous permet de construire une méthodologie appliquée de l’ANT: analyse bibliométrique (Publish Or Perish), text mining, clustering et analyse sémantique de la littérature scientifique et web du neuromarketing. A partir de ces résultats, nous construisons des cartographies, sous forme de graphes en réseau (Gephi) qui révèlent les interrelations et les associations entre acteurs, traces et controverses autour du neuromarketing
Our research explores the comparative development of neuromarketing between the United States and France. We start by analyzing the literature on neuromarketing. We use as theoretical and methodological framework the Actor Network Theory (ANT) (in the wake of the work of Bruno Latour and Michel Callon). We show how “human and non-human” entities (“actants”): actor-network, traces (publications) and controversies form the pillars of a new discipline such as the neuromarketing. Our hybrid approach “qualitative-quantitative” allows us to build an applied methodology of the ANT: bibliometric analysis (Publish Or Perish), text mining, clustering and semantic analysis of the scientific literature and web of the neuromarketing. From these results, we build data visualizations, mapping of network graphs (Gephi) that reveal the interrelations and associations between actors, traces and controversies about neuromarketing

Ce travail porte sur l’analyse des parcours utilisateurs dans les bibliothèques numériques qui se caractérisent par des processus métier faiblement structurés. Nous allons avoir recours à la fouille de processus afin d’extraire les modèles à partir de ces parcours. Les modèles découverts permettront aux concepteurs de ces systèmes de répondre d’une manière plus efficace aux besoins des utilisateurs et à leurs présenter un ensemble de recommandations. Dans cette thèse, nous avons généré des modèles à partir des traces de navigation des utilisateurs du portail web de la bibliothèque nationale de France, Gallica. Dans un premier temps, nous avons adapté ces traces sous un format compatible avec les algorithmes de fouille de processus. Dans un second temps, nous avons regroupé ces traces. L’originalité de notre apport porte sur le regroupement des parcours similaires, en tenant compte des caractéristiques existantes dans les traces, afin d’éviter la génération des modèles complexes, souvent non exploitables, à partir de telles traces volumineuse set non structurées. Enfin, nous avons validé notre méthode sur deux jeux de données simulé et réel. Nous avons comparé notre méthode à deux méthodes inspirées des travaux existants et les résultats montrent que notre méthode surpasse celles existantes sur les deux jeux de données à la fois dans le regroupement et la modélisation
This research focuses on extracting users’ journeys in a digital library characterized by weakly structured business processes. In this thesis, we investigate whether it is possible to model user journeys using process mining. The discovered models allow system designers to respond more efficiently to users’ needs and to present them with a set of recommendations. For our study, we have chosen to extract the users’ journey models of the digital library Gallica, based on real traces generated by their users. First, we adapt these browsing traces in a well-defined format compatible with process mining techniques. The originality of our contribution concerns the grouping of similar paths, considering the existing characteristics in the traces, to avoid the generation of complex models, often not exploitable, from such voluminous and unstructured traces. Finally, we validate our method on two simulated and real data sets. We compare our method to two other methods inspired by existing works. The results show that our method outper forms the existing ones on both datasets in clustering and modeling

ACTAR TPC (ACtive TARget and Time Projection Chamber) est une cible active de nouvelle génération construite au GANIL (Grand Accélérateur d'Ions Lourds). Les cibles actives sont des cibles gazeuses où le gaz permet de détecter le passage de particules chargées selon le principe des chambres à projection temporelle (TPC). La TPC d'ACTAR est formée d'une anode segmentée de 16384 pixels carrés de 2 mm de côté. La haute densité de voies est gérée par le système électronique GET (General Electronics for TPCs). Ce système digitalise également les signaux sur un intervalle de temps donné, pour une reconstruction 3D complète des évènements. Un démonstrateur huit fois plus petit a d'abord été construit pour vérifier le fonctionnement de l’électronique et la conception mécanique. La finalisation d’ACTAR TPC s’est basée sur les résultats du démonstrateur, qui a été testé avec des faisceaux de 6Li, de 24Mg et de 58Ni. Le commissioning d'ACTAR TPC a été effectué dans le cas de la diffusion résonante sur cible de protons avec des faisceaux de 18O et de 20Ne.Un algorithme de reconstruction des traces mesurées dans la TPC permet d'en extraire les angles et les énergies des ions impliqués dans les réactions. Les résultats sont comparés à des données connues pour déterminer les performances du système de détection. Les résolutions obtenues sur le commissioning à partir de calculs en matrice R valident l'utilisation d'ACTAR TPC pour de futures expériences. Par ailleurs, la diffusion résonante 6Li + 4He réalisée avec le démonstrateur a permis d'étudier les états d’agrégat alpha dans le 10B. Deux résonances à 8.58 MeV et 9.52 MeV sont observées pour la première fois en diffusion élastique dans cette voie de réaction
ACTAR TPC (ACtive TARget and Time Projection Chamber) is a next generation active target that was designed and built at GANIL (Grand Accélérateur d'Ions Lourds). Active targets are gaseous targets in which the gas is also used to track charged particles following the principles of time projection chambers (TPC). The TPC of ACTAR has a segmented anode of 16384 2 mm side square pixels. The high density of pixels is processed using the GET (General Electronics for TPCs) electronic system. This system also digitizes the signals over a time interval, enabling a full 3D event reconstruction. An eight time smaller demonstrator was first built to verify the electronics operation and the mechanical design. ACTAR TPC's final design was based on results obtained with the demonstrator which was tested using 6Li, 24Mg and 58Ni beams. The commissioning of ACTAR TPC was then carried out for the case of resonant scattering on a proton target using 18O and 20Ne beams. A track reconstruction algorithm is used to extract the angles and energies of the ions involved in the reactions. Results are compared to previous data to determine the detection system performances. Comparing the commissioning data with R matrix calculations, excitation functions resolutions in different cases are obtained. The use of ACTAR TPC is validated for future experiments. Furthermore, alpha clustering was studied in 10B through the resonant scattering 6Li + 4He, carried out with the demonstrator. Two resonances at 8.58 MeV and 9.52 MeV are observed for the first time in elastic scattering with this reaction channel

En utilisant des données haute-fréquence inédites, cette thèse étudie trois types de regroupements (“clusters”) présents dans le marché des changes: la concentration d'ordres sur certains prix, la concentration des transactions dans le temps et l'existence de groupes d'investisseurs prenant les mêmes décisions. Nous commençons par étudier les propriétés statistiques du carnet d'ordres EBS pour les paires de devises EUR/USD et USD/JPY et l'impact d'une réduction de la taille du tick sur sa dynamique. Une grande part des ordres limites est encore placée sur les anciens prix autorisés, entraînant l'apparition de prix-barrières, où figurent les meilleures limites la plupart du temps. Cet effet de congestion se retrouve dans la forme moyenne du carnet où des pics sont présents aux distances entières. Nous montrons que cette concentration des prix est causée par les traders manuels qui se refusent d’utiliser la nouvelle résolution de prix. Les traders automatiques prennent facilement la priorité, en postant des ordres limites un tick devant les pics de volume.Nous soulevons ensuite la question de l'aptitude des processus de Hawkes à rendre compte de la dynamique du marché. Nous analysons la précision de tels processus à mesure que l'intervalle de calibration est augmenté. Différent noyaux construits à partir de sommes d'exponentielles sont systématiquement comparés. Le marché FX qui ne ferme jamais est particulièrement adapté pour notre but, car il permet d’éviter les complications dues à la fermeture nocturne des marchés actions. Nous trouvons que la modélisation est valide selon les trois tests statistiques, si un noyau à deux exponentielles est utilisé pour fitter une heure, et deux ou trois pour une journée complète. Sur de plus longues périodes la modélisation est systématiquement rejetée par les tests à cause de la non-stationnarité du processus endogène. Les échelles de temps d'auto-excitation estimées sont relativement courtes et le facteur d'endogénéité est élevé mais sous-critique autour de 0.8. La majorité des modèles à agents suppose implicitement que les agents interagissent à travers du prix des actifs et des volumes échangés. Certains utilisent explicitement un réseau d'interaction entre traders, sur lequel des rumeurs se propagent, d'autres, un réseau qui représente des groupes prenant des décisions communes. Contrairement à d'autres types de données, de tels réseaux, s'ils existent, sont nécessairement implicites, ce qui rend leur détection compliquée. Nous étudions les transactions des clients de deux fournisseur de liquidités sur plusieurs années. En supposant que les liens entre agents sont déterminés par la synchronisation de leur activité ou inactivité, nous montrons que des réseaux d'interactions existent. De plus, nous trouvons que l'activité de certains agents entraîne systématiquement l’activité d'autres agents, définissant ainsi des relations de type “lead-lag” entre les agents. Cela implique que le flux des clients est prévisible, ce que nous vérifions à l'aide d'une méthode sophistiquée d'apprentissage statistique
The aim of this thesis is to study three types of clustering in foreign exchange markets, namely in price, trades arrivals and investors decisions. We investigate the statistical properties of the EBS order book for the EUR/USD and USD/JPY currency pairs and the impact of a ten-fold tick size reduction on its dynamics. A large fraction of limit orders are still placed right at or halfway between the old allowed prices. This generates price barriers where the best quotes lie for much of the time, which causes the emergence of distinct peaks in the average shape of the book at round distances. Furthermore, we argue that this clustering is mainly due to manual traders who remained set to the old price resolution. Automatic traders easily take price priority by submitting limit orders one tick ahead of clusters, as shown by the prominence of buy (sell) limit orders posted with rightmost digit one (nine).The clustering of trades arrivals is well-known in financial markets and Hawkes processes are particularly suited to describe this phenomenon. We raise the question of what part of market dynamics Hawkes processes are able to account for exactly. We document the accuracy of such processes as one varies the time interval of calibration and compare the performance of various types of kernels made up of sums of exponentials. Because of their around-the-clock opening times, FX markets are ideally suited to our aim as they allow us to avoid the complications of the long daily overnight closures of equity markets. One can achieve statistical significance according to three simultaneous tests provided that one uses kernels with two exponentials for fitting an hour at a time, and two or three exponentials for full days, while longer periods could not be fitted within statistical satisfaction because of the non-stationarity of the endogenous process. Fitted timescales are relatively short and endogeneity factor is high but sub-critical at about 0.8.Most agent-based models of financial markets implicitly assume that the agents interact through asset prices and exchanged volumes. Some of them add an explicit trader-trader interaction network on which rumors propagate or that encode groups that take common decisions. Contrarily to other types of data, such networks, if they exist, are necessarily implicit, which makes their determination a more challenging task. We analyze transaction data of all the clients of two liquidity providers, encompassing several years of trading. By assuming that the links between agents are determined by systematic simultaneous activity or inactivity, we show that interaction networks do exist. In addition, we find that the (in)activity of some agents systematically triggers the (in)activity of other traders, defining lead-lag relationships between the agents. This implies that the global investment flux is predictable, which we check by using sophisticated machine learning methods

The trace element content of clinkers (and possibly of cements) can be used to identify the manufacturing factory. The Mg, Sr, Ba, Mn, Ti, Zr, Zn and V content of clinkers give detailed information for the determination of the origin of clinkers produced in different factories. However, for the analysis of such complex data there is a need for algorithmic tools for the visualization and clustering of the samples. This paper proposes a new approach for this purpose. The analytical data are transformed into a twodimensional latent space by factor analysis (probabilistic principal component analysis) and dendograms are constructed for cluster formation. The classification of South African clinkers is used as an illustrative example for the approach.

Traffic simulators are effective tools to support decisions in urban planning systems, to identify criticalities, to observe emerging behaviours in road networks and to configure road infrastructures, such as road side units and traffic lights. Clearly the more realistic the simulator the more precise the insight provided to decision makers. This paper provides a first step toward the design and calibration of traffic micro-simulator to produce realistic behaviour. The long term idea is to collect and analyse real traffic traces collecting vehicular information, to cluster them in groups representing similar driving behaviours and then to extract from these clusters relevant parameters to tune the microsimulator. In this paper we have run controlled experiments where traffic traces have been synthetized to obtain different driving styles, so that the effectiveness of the clustering algorithm could be checked on known labels. We describe the overall methodology and the results already achieved on the controlled experiment, showing the clusters obtained and reporting guidelines for future experiments.

Abstract The trace element content of clinkers (and possibly of cements) can be used to identify the manufacturing factory. The Mg, Sr, Ba, Mn, Ti, Zr, Zn and V content of clinkers give detailed information for the determination of the origin of clinkers produced in different factories. However, for the analysis of such complex data there is a need for algorithmic tools for the visualization and clustering of the samples. This paper proposes a new approach for this purpose. The analytical data are transformed into a twodimensional latent space by factor analysis (probabilistic principal component analysis) and dendograms are constructed for cluster formation. The classification of South African clinkers is used as an illustrative example for the approach.

Diffusion tensor imaging (DTI) is a structural magnetic resonance imaging (MRI) technique to record incoherent motion of water molecules and has been used to detect micro structural white matter alterations in clinical studies to explore certain brain disorders. A variety of DTI based techniques for detecting brain disorders and facilitating clinical group analysis have been developed in the past few years. However, there are two crucial issues that have great impacts on the performance of those algorithms. One is that brain neural pathways appear in complicated 3D structures which are inappropriate and inaccurate to be approximated by simple 2D structures, while the other involves the computational efficiency in classifying white matter tracts. The first key area that this dissertation focuses on is to implement a novel computing scheme for estimating regional white matter alterations along neural pathways in 3D space. The mechanism of the proposed method relies on white matter tractography and geodesic distance mapping. We propose a mask scheme to overcome the difficulty to reconstruct thin tract bundles. Real DTI data are employed to demonstrate the performance of the pro- posed technique. Experimental results show that the proposed method bears great potential to provide a sensitive approach for determining the white matter integrity in human brain. Another core objective of this work is to develop a class of new modeling and clustering techniques with improved performance and noise resistance for separating reconstructed white matter tracts to facilitate clinical group analysis. Different strategies are presented to handle different scenarios. For whole brain tractography reconstructed white matter tracts, a Fourier descriptor model and a clustering algorithm based on multivariate Gaussian mixture model and expectation maximization are proposed. Outliers are easily handled in this framework. Real DTI data experimental results show that the proposed algorithm is relatively effective and may offer an alternative for existing white matter fiber clustering methods. For a small amount of white matter fibers, a modeling and clustering algorithm with the capability of handling white matter fibers with unequal length and sharing no common starting region is also proposed and evaluated with real DTI data.

Le recalage des images du cerveau vise à réduire la variabilité anatomique entre les differentes sujets, et à créer un espace commun pour l'analyse de groupe. Les approches multi-modales essaient de minimiser les variations de forme du cortex et des structures internes telles que des faisceaux de fibres nerveuses. Ces approches nécessitent une identification préalable de ces structures, ce qui s'avère une tâche difficile en l'absence d'un atlas complet de référence. Nous proposons une extension de l'algorithme de recalage difféomorphe des Démons pour recaler conjointement des images et des faisceaux de fibres. Dans cette thèse, nous analysons différentes représentations des faisceaux de fibres comme une séquence de points, un nuage de points, les courants et les mesures. Différentes distances sont analysées et étudiées dans l'algorithme de recalage. Pour simplifier la représentation de la matière blanche nous utilisons et étendons les algorithmes de classification existants. En étendant le recalage d'images afin d'ajouter des descripteurs de la géométrie des fibres nerveuses, nous espérons améliorer les futures analyses concernant les matières grise et blanche. Nous avons démontré l'efficacité de notre algorithme en recalant conjointement des images anatomiques pondérées en T1 et des faisceaux de fibres. Nous avons comparé nos résultats à des approches concurrentes, l'une multimodale s'appuyant sur l'anisotropie fractionnaire et la pondération T1, l'autre sur les tenseurs de diffusion, et obtenu de meilleures performances à l'aide de notre algorithme. Enfin, nous mettons en évidence sur des études de groupe en IRMf que notre méthodologie et notre implémentation apportent un gain en sensibilité de détection des activations cérébrales
Brain image registration aims at reducing anatomical variability across subjects to create a common space for group analysis. Multi-modal approaches intend to minimize cortex shape variations along with internal structures, such as fiber bundles. These approaches require prior identification of the structures, which remains a challenging task in the absence of a complete reference atlas. We propose an extension of the Diffeomorphic Demons image registration to jointly register images and fiber bundles. In this thesis we analyze differents representations of the fiber bundles such as ordered points, clouds of points, Currents and Measures. Different distances are analyzed and implemented into the registration algorithm. To simplify white matter representation we also analyze, use and extend existing clustering algorithms. By extending the image registration to include geometric fiber bundles descriptors we hope to improve future analyses regarding both, grey and white matter. We demonstrate the efficacy of our algorithm by registering simultaneously T1 images and fiber bundles and compare results with a multi-modal T1+Fractional Anisotropy (FA) and a tensor-based registration algorithms and obtain superior performance with our approach. We provide preliminary evidence that our implementation improves the sensitivity of activation detection in fMRI group studies

The accelerated expansion of the Universe and the nature of the Dark Energy are still open questions in cosmology. One of the most powerful ways to investigate these issues is to map the large-scale structure of the Universe, to constrain its expansion history and growth of structures. In particular, baryon acoustic oscillations (BAO) occurred at recombination make a peak in the correlation function of galaxies at the characteristic scale of the sound horizon (a sufficiently large scale to “protect” the signal from strong non-linearities), or alternatively a series of oscillations in the power spectrum. Since the sound horizon can be estimated with a great precision from the position of the first peak in the angular power spectrum of the Cosmic Microwave Background (which has the same physical origin of BAO, oscillations of the baryons-photons plasma), the BAO peak in the correlation function can be used as a standard ruler, providing paramount cosmological information. The aim of this thesis is to systematically test and possibly improve the state-of- the-art statistical methods to model the BAO peak, taking into account the non-linear evolution of matter overdensities, redshift-space distortions and the bias of cosmic tracers. To do that, we analyse mock samples of galaxies, quasars and galaxy clusters extracted from one of the largest available cosmological hydrodynamical simulations. We extract cosmological constraints from the BAO peak through different statistical tools in the redshift range 0.2 < z < 2. Although the BAO peak is at large scales, non-linear growth and galaxy peculiar velocities make the BAO signal smoothed and broader with respect to linear predictions, especially at low redshifts. A possible method to overcome these issues is the so-called reconstruction of the density field: one of the primary goals of this work is to implement a reconstruction method, to check its performances as a function of sample selections and redshift.

Notre travail porte sur l'extraction de connaissances sur des langages graphiques dont les symboles sont a priori inconnus. Nous formons l'hypothèse que l'observation d'une grande quantité de documents doit permettre de découvrir les symboles composant l'alphabet du langage considéré. La difficulté du problème réside dans la nature bidimensionnelle et manuscrite des langages graphiques étudiés. Nous nous plaçons dans le cadre de tracés en-ligne produit par des interfaces de saisie de type écrans tactiles, tableaux interactifs ou stylos électroniques. Le signal disponible est alors une trajectoire échantillonnée produisant une séquence de traits, eux-mêmes composés d'une séquence de points. Un symbole, élément de base de l'alphabet du langage, est donc composé d'un ensemble de traits possédant des propriétés structurelles et relationnelles spécifiques. L'extraction des symboles est réalisée par la découverte de sous-graphes répétitifs dans un graphe global modélisant les traits (noeuds) et leur relations spatiales (arcs) de l'ensemble des documents. Le principe de description de longueur minimum (MDL : Minimum Description Length) est mis en oeuvre pour choisir les meilleurs représentants du lexique des symboles. Ces travaux ont été validés sur deux bases expérimentales. La première est une base d'expressions mathématiques simples, la seconde représente des graphiques de type organigramme. Sur ces bases, nous pouvons évaluer la qualité des symboles extraits et comparer à la vérité terrain. Enfin, nous nous sommes intéressés à la réduction de la tâche d'annotation d'une base en considérant à la fois les problématiques de segmentation et d'étiquetage des différents traits.

The present Ph.D. thesis is contextualized within the Muon Portal project, a project dedicated to the creation of a tomograph for the control and scanning of containers at the border in order to reveal smuggled fissile material by means of the cosmic muons scattering. This work aims to extend and consolidate the research in the field of muon tomography in the context of applied physics. The main purpose of the thesis is to investigate new techniques for reconstruction of muon tracks within the detector and new approaches to the analysis of data from muon tomography for the automatic objects recognition and the 3D visualization, thus making possi- ble the realization of a tomography of the entire container. The research work was divided into different phases, described in this thesis document: from a prelimi- nary speculative study of the state of the art on the tracking issue and on the tracks reconstruction algorithms, to the study on the Muon Portal detector performance in the case of particle tracking at low and high multiplicity. A substantial part of the work was devoted to the study of different image reconstruction techniques based on the POCA algorithm (Point of Closest Approach) and the iterative EM-LM algorithm (Expectation-Maximization). In addition, more advanced methods for the tracks reconstruction and visualization, such as data-mining techniques and clustering algorithms have been the subject of the research and development ac- tivity which has culminated in the development of an unsupervised multiphase clustering algorithm (modified-Friends-of-Friends) for the muon tomography data analysis.

Parkinson’s disease patients present with the highest risk of dementia development. The thalamus, integral to several functions and behaviours is involved in the pathophysiology of Parkinson’s disease. The aim of this thesis was to determine if anatomical abnormalities in the thalamus are associated with the development of dementia in Parkinson’s disease. We examined the thalamus using macro and microstructural techniques and the white matter pathways that connect the thalamus with areas of the surrounding cortex using diffusion tensor imaging (DTI) based tractography. T1-weighted magnetic resonance and DT images were collected in 56 Parkinson’s disease patients with no cognitive impairment, 19 patients with mild cognitive impairment, 17 patients with dementia and 25 healthy individuals who acted as control subjects. An established automated segmentation procedure (FIRST FSL) was used to delineate the thalamus and a modified k-means clustering algorithm applied to segment the thalamus into clusters assumed to represent thalamic nuclei. Fibre tracts were determined using DTI probabilistic tracking methods available in FIRST. Microstructural integrity was quantified by fractional anisotropy and mean diffusivity (MD) DTI measures. Results show that microstructural measures of thalamic integrity are more sensitive to cognitive dysfunction in PD than macrostructural measures. For the first time we showed a progressive worsening of cellular integrity (MD) in the groups who had greater levels of cognitive dysfunction. Thalamic degeneration was regionally specific and most advanced in the limbic thalamic nuclei which influenced executive function and attention, areas of cognition that are known to be affected in the earliest stages of PD. The integrity of the fibre tracts corresponding to these thalamic regions was also compromised. Degeneration of fibre tracts was most evident in the dementia group, indicating that they may be more protected against Lewy pathology than the nuclei of the thalamus. Our findings confirm previous histological, animal and lesion studies and provide a reliable estimate of cortical degeneration in PD that can be applied non-invasively and in vivo. A longitudinal study is needed to monitor the progression of cognitive decline in PD but we have provided the basis for further investigation into the predictive validity of thalamic degeneration for cognitive dysfunction. In the future, the microstructural changes of the thalamus could be used as biomarkers for the identification of individuals with a higher risk for dementia development and for the longitudinal monitoring of any interventions into cognitive decline.

Analyses of nuclear reactor safety have increasingly required coupling of full three dimensional neutron kinetics (NK) core models with system transient thermal–hydraulics (TH) codes.  In order to produce results within a reasonable computing time, the coupled codes use two different spatial description of the reactor core.  The TH code uses few, typically 5 to 20 TH channels, which represent the core.  The NK code uses explicit one node for each fuel assembly.  Therefore, a spatial mapping of a coarse grid TH and a fine grid NK domain is necessary.  However, improper mappings may result in loss of valuable information, thus causing inaccurate prediction of safety parameters. The purpose of this thesis is to study the effectiveness of spatial coupling (channel refinement and spatial mapping) and develop recommendations for NK/TH mapping in simulation of safety transients.  Additionally, sensitivity of stability (measured by Decay Ratio and Frequency) to the different types of mapping schemes, is analyzed against OECD/NEA Ringhals–1 Stability Benchmark data. The research methodology consists of spatial coupling convergence study, by increasing the number of TH channels and varying mapping approaches, up to and including the reference case.  The reference case consists of one-to-one mapping: one TH channel per one fuel assembly.  The comparisons of the results are done for steady–state and transient results.  In this thesis mapping (spatial coupling) definition is formed and all the existing mapping approaches were gathered, analyzed and presented.  Additionally, to increase the efficiency and applicability of spatial mapping convergence, a new mapping methodology has been proposed.  The new mapping approach is based on hierarchical clustering method; the method of unsupervised learning that is adopted by many researchers in many different scientific fields, thanks to its flexibility and robustness.  The proposed new mapping method turns out to be very successful for spatial coupling problem and can be fully automatized allowing for significant time reduction in mapping convergence study. The steady–state results obtained from three different plant models for all the investigated cases are presented.  All models achieved well converged steady–state and local parameters were compared and it was concluded that solid basis for further transient analysis was found.  Analyzing the mapping performance, the best predictions for steady–state conditions are the mappings that include the power peaking factor feature alone or with any combination of other features.  Additionally it is of value to keep the core symmetry (symmetry feature).  The big part of this research is devoted to transient analysis.  The selection of transients was done such that it covers a wide range of transients and gathered knowledge may be used for other types of transients.  As a representative of a local perturbation, Control Rod Drop Accident was chosen.  A specially prepared Feedwater Transient was investigated as a regional perturbation and a Turbine Trip is an example of a global one.  In the case of local perturbation, it has been found that a number of TH channels is less important than the type of mapping, so a high number of TH channels does not guarantee improved results.  To avoid unnecessary averaging and to obtain the best prediction, hot channel and core zone where accident happens should be always separated from the rest.  The best performance is achieved with mapping according power peaking factors, and therefore this one is recommended for such type of perturbation. The regional perturbation has been found to be more challenging than the others.  This kind of perturbation is strongly dependent on mapping type that affects the power increase rate, SCRAM time, onset of instability, development of limit cycle, etc.  It has been also concluded that a special effort is needed for input model preparation.   In contrast to the regional perturbation, the global perturbation is found to be the least demanding transient.  Here, the number of TH channels and type of mapping do not have significant impact on average plant behaviour – general plant response is always well recreated.  A special effort has also been paid to investigate the core stability performance, in both global and regional mode.  It has been found that in case of unstable cores, a low number of TH channels significantly suppresses the instability.  For these cases number of TH channels is very important and therefore at least half of the core has to be modeled to have a confidence in predicted DR and FR.  In case of regional instability in order to get correct performance of out-of-phase oscillations, it is recommended to use full-scale model.  If this is not possible, the mapping which is a mixture of 1st power mode and power peaking factors, should be used. The general conclusions and recommendations are summarized at the end of this thesis.  Development of these recommendations was one of the purposes of this investigation and they should be taken into consideration while designing new coupled TH/NK models and choosing mapping strategy for a new transient analysis.

QC 20130516

Les rayonnements ionisants sont connus pour induire des dommages critiques au sein de la matière biologique et spécialement au sein de l’ADN. Parmi ces dommages, les cassures doubles brins de l’ADN (DSB) sont considérées comme les principales responsables des effets létaux des rayonnements. Comprendre et prédire comment ces cassures sont créées et réparées dans les noyaux cellulaires demeure un défi dans la recherche en radiobiologie. Ce travail s’inscrit dans ce contexte, dans la modélisation des cassures double brin de l’ADN (DSB) à partir des dépôts d’énergie créés par l’irradiation au niveau intracellulaire. Le détail topologique au niveau nanométrique des dépôts d’énergie nécessaire à ce travail est obtenu par modélisation Monte Carlo à l’aide du code Geant4 et, en particulier son extension Geant4-DNA pour des processus à très faible énergie. Les dommages étudiés étant ceux localisés dans l’ADN, le premier objectif de ce travail a été de réaliser une géométrie détaillée de celui-ci afin de l’implémenter dans les calculs Monte Carlo. Deux types de noyaux cellulaires, représentant un fibroblaste et un endothélium, ont été décrits afin d’évaluer l’influence de la densité d’ADN dans les résultats sur la topologie des dépôts pouvant donner lieux à des cassures de la molécule. Cette géométrie nous permet d’effectuer une première sélection des dépôts d’énergie pouvant contribuer aux cassures car situées sur la chaîne sucre-phosphate. Ces dépôts sont ensuite analysés à l’aide d’un algorithme de clustérisation de manière à les regrouper sous forme d’agrégats afin d’étudier leur localisation et complexité. Néanmoins, dans cette étude, seule les interactions physiques entre les rayonnements ionisants et la cible sont modélisées, il n’est donc pas possible d’obtenir un nombre absolu de cassures de brins car cette modélisation n’inclue pas l’étape de création et de transport des radicaux libres pouvant donner lieu à des dommages indirects. Ainsi, le but de ce travail était d’évaluer la dépendance relative des dommages radio-induits directs avec la densité d’ADN, la qualité du rayonnement, la morphologie du noyau ou encore la condensation de la chromatine. Les différentes modélisations réalisées ont permis de quantifier l’influence de ces différents paramètres dans le nombre et la complexité des dommages directs induits dans l’ADN, pouvant ensuite contribuer aux effets tardifs sur le devenir cellulaire
Ionizing radiations are known to induce critical damages on biological matter and especially on DNA. Among these damages, DNA double strand breaks (DSB) are considered as key precursor of lethal effects of ionizing radiations. Understand and predict how DNA double and simple strand breaks are created by ionising radiation and repaired in cell nucleus is nowadays a major challenge in radiobiology research. This work presents the results on the simulation of the DNA double strand breaks produced from the energy deposited by the irradiation at the intracellular level. At the nanometric scale, the only method to accurately simulate the topological details of energy deposited on the biological matter is the use of Monte Carlo codes. In this work, we used the Geant4 Monte Carlo code and, in particular, the low energy electromagnetic package extensions, referred as Geant4-DNA processes.In order to evaluate DNA radio-induced damages, the first objective of this work consisted in implementing a detailed geometry of the DNA on the Monte Carlo simulations. Two types of cell nuclei, representing a fibroblast and an endothelium, were described in order to evaluate the influence of the DNA density on the topology of the energy deposits contributing to strand breaks. Indeed, the implemented geometry allows the selection of energy transfer points that can lead to strand breaks because they are located on the backbone. Then, these energy transfer points were analysed with a clustering algorithm in order to reveal groups of aggregates and to study their location and complexity.In this work, only the physical interactions of ionizing radiations are simulated. Thus, it is not possible to achieve an absolute number of strand breaks as the creation and transportation of radical species which could lead to indirect DNA damages is not included. Nevertheless, the aim of this work was to evaluate the relative dependence of direct DNA damages with the DNA density, radiation quality, cell nuclei morphology or also chromatin condensation. The results presented in this work have allowed the quantification of the influence of these different parameters in the number and complexity of directs DNA damages which can then contribute to the late effects on cell fate

La structure et l'organisation de la substance blanche du cerveau humain ne sont pas encore complètement connues. L'Imagerie par Résonance Magnétique de diffusion (IRMd) offre une approche unique pour étudier in vivo la structure des tissus cérébraux, permettant la reconstruction non invasive des trajectoires des faisceaux de fibres du cerveau en utilisant la tractographie. Aujourd'hui, les techniques récentes d'IRMd avec haute résolution angulaire (HARDI) ont largement amélioré la qualité de la tractographie par rapport à l'imagerie du tenseur de diffusion standard (DTI). Toutefois, les jeux de données de tractographie résultant sont très complexes et comprennent des millions de fibres, ce qui nécessite une nouvelle génération de méthodes d'analyse. Au-delà de la cartographie des principales voies de la substance blanche, cette nouvelle technologie ouvre la voie à l'étude des faisceaux d'association courts, qui ont rarement été étudiés avant et qui sont au centre de cette thèse. L'objectif est d'inférer un atlas des faisceaux de fibres du cerveau humain et une méthode qui permet le mappage de cet atlas à tout nouveau cerveau.Afin de surmonter la limitation induite par la taille et la complexité des jeux de données de tractographie, nous proposons une stratégie à deux niveaux, qui enchaîne des regroupements de fibres intra- et inter-sujet. Le premier niveau, un regroupement intra-sujet, est composé par plusieurs étapes qui effectuent un regroupement hiérarchique et robuste des fibres issues de la tractographie, pouvant traiter des jeux de données contenant des millions de fibres. Le résultat final est un ensemble de quelques milliers de faisceaux de fibres homogènes représentant la structure du jeu de données de tractographie dans sa totalité. Cette représentation simplifiée de la substance blanche peut être utilisée par plusieurs études sur la structure des faisceaux individuels ou des analyses de groupe. La robustesse et le coût de l'extensibilité de la méthode sont vérifiés à l'aide de jeux de fibres simulés. Le deuxième niveau, un regroupement inter-sujet, rassemble les faisceaux obtenus dans le premier niveau pour une population de sujets et effectue un regroupement après normalisation spatiale. Il produit en sortie un modèle composé d'une liste de faisceaux de fibres génériques qui peuvent être détectés dans la plupart de la population. Une validation avec des jeux de données simulées est appliqué afin d'étudier le comportement du regroupement inter-sujet sur une population de sujets alignés avec une transformation affine. La méthode a été appliquée aux jeux de fibres calculés à partir des données HARDI de douze cerveaux adultes. Un nouveau atlas des faisceaux HARDI multi-sujet, qui représente la variabilité de la forme et la position des faisceaux à travers les sujets, a été ainsi inféré. L'atlas comprend 36 faisceaux de la substance blanche profonde, dont certains représentent quelques subdivisions des faisceaux connus, et 94 faisceaux d'association courts de la substance blanche superficielle. Enfin, nous proposons une méthode de segmentation automatique de mappage de cet atlas à tout nouveau sujet.

Dissertação de Mestrado em Engenharia Informática apresentada à Faculdade de Ciências e Tecnologia
Location-Based Services are increasingly present in our daily lives. However, regardless of the benefits that these services offer to users, the shared data are not always and only used for the initial purpose. These data can be made public or sold, for example, for commercial purposes. The fact that location data contain information that can reveal the person’s identity, routines and habits, raises serious privacy concerns. In order to respond to this problem, there are privacy-preserving mechanisms, namely, for obfuscation and for anonymization of data. However, the correlation between location reports, which can potentially be used by an adversary to estimate the position of the user, has been underlooked in privacy protection. The aim of this thesis is to develop a user-centric Location Privacy-Preserving Mechanism, that is, a mechanism that protects privacy of a user at collection time. In addition, it is intended to protect the users not only against single reports, but also over time, against continuous reports. In this latter scenario, we intent to develop a protection mechanism that is suitable to different frequency of updates and/or to the correlation between reports as to mitigate possible privacy violations that advent from exploring these intrinsic characteristics of location data. Towards this end, we started by evaluating the impact of the frequency of updates on location privacy. For that, we implemented a state-of-the-art tracking attack that allows us to assess the effect of the frequency of updates by estimating the exact user locations. According to the performed analysis, we developed a new mechanism based on geo-indistinguishability that creates obfuscation clusters to aggregate closer locations. This developed mechanism is designated clustering geo-indistinguishability. To evaluate the utility of the mechanism, we resorted to a real use-case based on geofencing. Lastly, the evaluation of the mechanism enables us to conclude that it safeguards the level of privacy and the utility of continuous reports of location data, in a way that it can still be used for the purpose of a service.
Os serviços baseados em localização estão cada vez mais presentes no nosso quotidiano. No entanto, apesar do benefício que estes serviços oferecem aos utilizadores, os dados partilhados nem sempre são usados apenas com o propósito inicial. Estes dados podem ser tornados públicos ou vendidos, por exemplo, para fins comerciais. O facto dos dados de localização conterem informações passíveis de revelar a identidade, as rotinas e os hábitos de uma pessoa, levantam sérias preocupações de privacidade. Para dar resposta a este desiderato, existem mecanismos de preservação de privacidade, nomeadamente, de ofuscação e anonimização dos dados. Contudo, a correlação entre os dados de localização partilhados, que pode ser usada por um adversário para estimar a posição de um utilizador, tem sido negligenciada na proteção da privacidade. O objetivo desta tese é desenvolver um mecanismo de preservação de privacidade de localização centrado no utilizador, isto é, um mecanismo que proteja os utilizadores no momento da partilha de dados. Para além disso, pretende-se proteger o utilizador não só quando este reporta localizações únicas, mas também ao longo do tempo, isto é, quando reporta localizações de modo contínuo. Neste último cenário, pretendemos desenvolver um mecanismo de proteção que seja adequado a diferentes frequências de atualização de localização e/ou à correlação existente entre as localizações partilhadas, de modo a mitigar possíveis violações de privacidade que advenham da exploração destas características intrínsecas dos dados de localização. Neste sentido, começámos por avaliar o impacto da frequência na privacidade de localização. Para tal, implementámos um ataque considerado estado da arte que permite localizar o utilizador ao longo do tempo e do espaço, viabilizando a avaliação do efeito da frequência através da estimação da localização exata do utilizador. De acordo com a análise efetuada, desenvolvemos um mecanismo novo baseado em geo-indistinguishability que cria áreas de ofuscação para agregar localizações próximas. O mecanismo desenvolvido é designado clustering geo-indistinguisahbility. Para avaliar a utilidade do mecanismo, utilizámos um caso de uso real baseado em geofencing. Por fim, a avaliação do mecanismo permitiu-nos concluir que este salvaguarda o nível de privacidade e a utilidade dos dados, de tal modo que continuam a poder ser usados para o propósito do serviço.
Outro - Este trabalho é suportado pelos projetos SWING2 (PTDC/EEI-TEL/3684/2014) e Mobiwise (P2020 SAICTPAC/001/2015), financiado pelos Fundos Europeus Estruturais e de Investimento (FEEI) Europeus através do Programa Operacional Competitividade e Internacionalização - COMPETE 2020 e por Fundos Nacionais através da FCT - Fundação para a Ciência e a Tecnologia no âmbito do projeto POCI-01-0145-FEDER-016753, e pelo Fundo Europeu de Desenvolvimento Regional.

碩士
國立交通大學
資訊科學與工程研究所
100
Although a large number of malicious programs are created every day, most of them mutate from existing ones. These mutant malware programs may seemingly appear differently, but actually act with similar behavior patterns. By clustering these malware programs into the same cluster, the malware analysis effort can be reduced significantly. In this paper, we propose a clustering approach to malware classification by comparing instruction trace similarity of binary programs being tested. We take advantage of dynamic analysis to trace malware instructions at runtime. Our method can discover malware in disguise by using techniques such as polymorphism or code injection. By tracing malware instructions, our scheme ensure that the detection mechanism cannot be circumvented or sabotaged by malicious API tampering. The taint technique we adopted can filter massive instructions created by normal system library as noise to the malware analysis. Collected instruction traces are then compared to measure their similarity so that the clustering can be performed. The results demonstrate that our system is able to cluster malware with similar codes, and can recognize new malware which is undetected by anti-virus tools.

碩士
國立勤益科技大學
電子工程系
105
Nowadays many pictures can be produced on the cloth by a sewing machine. However, there are some drawback must be solved, like too many colors. In addition, color down of the original pictures to a few colors will appear some fragmented objects. Therefore, this research proposes a Color reduction and eliminate bitty pixels (CREBP) algorithm. There are two parts in this algorithm: The first part is: when the picture is used by the image clustering, we exclude the edge and then let the edge recolored. The second part is: find out the fragmented objects and let it recolored again.

Master of Science
Department of Statistics
Michael Higgins
The amount of data generated from stock market trading is massive. For example, roughly 10 million trades are performed each day on the NASDAQ stock exchange. A significant proportion of these trades are made by high-frequency traders. These entities make on the order of thousands or more trades a day. However, the stock-market factors that drive the decisions of high-frequency traders are poorly understood. Recently, hybridized threshold clustering (HTC) has been proposed as a way of clustering large-to-massive datasets. In this report, we use three months of NASDAQ HFT data---a dataset containing information on all trades of 120 different stocks including identifiers on whether the buyer and/or seller were high-frequency traders---to investigate the trading patterns of high-frequency traders, and we explore the use of HTC to identify these patterns. We find that, while HTC can be successfully performed on the NASDAQ HFT dataset, the amount of information gleaned from this clustering is limited. Instead, we show that an understanding of the habits of high-frequency traders may be gained by looking at \textit{janky} trades---those in which the number of shares traded is not a multiple of 10. We demonstrate evidence that janky trades are more common for high-frequency traders. Additionally, we suggest that a large number of small, janky trades may help signal that a large trade will happen shortly afterward.