No description available.

Most work that is done within machine learning today uses statistical methods which assume that the data is identically and independently distributed. However, the problem domains that we face in the real world are often much more complicated and present both complex relational/logical parts as well as parts with uncertainty. Statistical relational learning (SRL) is a sub-field of machine learning and A.I. that tries to solve these limitations by combining both relational and statistical learning and has become a big research sector in recent years. This thesis will present SRL further and specifically introduce, test and review one of the implementations, namely Bayesian logic programs. / Idag används inom maskininlärning nästan alltid statistiska metoder som antar att datat för lärande är identiskt och oberoende distribuerat. Men de problemområden som vi står inför i den verkliga världen är ofta mycket mer komplicerade och har både komplexa relationella/logiska delar samt osäkerhet. Statistiskt relationslärande (SRL) är en del av maskininlärning och A.I. som försöker lösa dessa begränsningar genom att kombinera både relationer och statistiskt lärande och har på senare år blivit ett stort forskningsområde. Denna avhandling presenterar SRL mer i detalj och utreder och testar en specifik implementation, Bayesianska logikprogram.

Nos travaux abordent le problème de l'efficacité des techniques d'implémentation hautes performances du système de communication. Car si les principes généraux sont désormais bien connus, il n'en va pas toujours de même de leur application. Des résultats obtenus sur notre plate-forme expérimentale, nous tirons des conclusions quant aux aspects machine hôte, système, et protocoles. Pour l'aspect machine hôte nous montrons que la parallélisation des piles de communication est décevante et grandement limitée par les contentions d'origine système. Nous montrons également comment la technique ILP, destinée à limiter les accès mémoire, peut être intégrée à la méthode d'accès d'une pile TCP/IP. Les bénéfices de cette technique restent largement dépendants des conditions d'utilisation, de la nature de l'application, et des manipulations de données. Du point de vue système, l'étude de deux environnements d'exécution de protocoles opposés, BSD et Streams, a permis d'identifier leurs faiblesses, et notamment pourquoi l'environnement Streams actuel n'est pas adapté aux hautes performances. Nous en déduisons un ensemble de principes qu'un environnement performant se doit de respecter. L'architecture de la pile de communication est essentielle. Nous montrons qu'une architecture démultiplexée, avec des chemins de données directs entre applications et drivers réseaux, permet un excellent support du contrôle de flux local et des mécanismes systèmes de QoS. En revanche, les implémentations de niveau utilisateur souffrent de nombreuses limitations. Elles sont cependant indispensables à certaines techniques. Du point de vue protocoles, nous montrons que la présence d'options dans les en-têtes de paquets n'est pas contraire à l'obtention de bonnes performances. A la notion trop rigide d'en-têtes de taille fixe nous substituons celle d'en-tétes de taille prédictible. Enfin, il ressort deux notions clés de ces travaux, la simplicité et la flexibilité, dont dépendent les performances et les fonctionnalités du système de communication.

Embedded system designers face multiple challenges in fulfilling the runtime requirements of programs. Effective scheduling of programs is required to extract as much parallelism as possible. These scheduling algorithms must also improve speedup after instruction-set extensions have occurred. Scheduling of dynamic code at run time is made more difficult when the static components of the program are scheduled inefficiently. This research aims to optimize a program’s static code at compile time. This is achieved with four algorithms designed to schedule code at the task and instruction level. Additionally, the algorithms improve scheduling using instruction set extended code on symmetrical homogeneous multiprocessor systems. Using these algorithms, we achieve speedups up to 3.86X over sequential execution for a 4-issue 2-processor system, and show better performance than recent heuristic techniques for small programs. Finally, the algorithms generate speedup values for a 64-point FFT that are similar to the test runs.

Embedded system designers face multiple challenges in fulfilling the runtime requirements of programs. Effective scheduling of programs is required to extract as much parallelism as possible. These scheduling algorithms must also improve speedup after instruction-set extensions have occurred. Scheduling of dynamic code at run time is made more difficult when the static components of the program are scheduled inefficiently. This research aims to optimize a program’s static code at compile time. This is achieved with four algorithms designed to schedule code at the task and instruction level. Additionally, the algorithms improve scheduling using instruction set extended code on symmetrical homogeneous multiprocessor systems. Using these algorithms, we achieve speedups up to 3.86X over sequential execution for a 4-issue 2-processor system, and show better performance than recent heuristic techniques for small programs. Finally, the algorithms generate speedup values for a 64-point FFT that are similar to the test runs.

L’objectiu principal d’aquesta tesis doctoral ha estat estudiar l’estat del Se en sang i LCR de pacients amb errors congènits del metabolisme amb tractament dietètic i altres trastorns neuropediàtrics de diferent etiologia. En el primer treball ens plantejàvem l’establiment de valors de referència per a determinats elements traça en sang total. La necessitat venia de la inexistència d’estudis sobre l’estat d’aquests elements en poblacions pediàtriques afectades de diferents errors congènits del metabolisme. En el segon treball ens plantejàvem l’establiment de valors de referència per a seleni en LCR. La necessitat venia de la inexistència d’estudis sobre l’estat d’aquest element en poblacions pediàtriques i de la possible associació entre una alteració de l’estat de Seleni i algunes disfuncions neurològiques. Un cop establerts els valors de referència de Se en LCR, el tercer treball consistia en valorar l’estat de Se en pacients amb malalties neurològiques per intentar detectar alguna deficiència o excés de Se en els líquids d’aquests pacients. A part de valorar l’estat de Se, preteníem valorar també altres paràmetres bioquímiques com HVA, 5-MTHF, 5-HIAA, proteïnes, activitat de GPx i pterines, per tal de veure si les correlacions observades en l’anterior treball es mantenien. Finalment, l’últim treball es centrava en l’estudi d’un grup de pacients amb uns valors de Se en LCR permanentment elevats sense tenir-ne una explicació clara. Aquests pacients (que s’havien identificat en el tercer treball) estaven afectats de la síndrome de Kearns-Sayre (malaltia mitocondrial que es manifesta per deleció del DNA mitocondrial). En aquest nou treball ens plantejàvem l’estudi d’uns paràmetres bioquímics que ens permetessin detectar la disfunció del plexe coroïdal en pacients afectats d’aquesta síndrome. Així doncs i a títol de discussió, una de les nostres principals necessitats pel control regular dels pacients amb malalties metabòliques i tractament dietètic era la monitorització de diferents elements traça en diferents fluids biològics. Tradicionalment s’havia descrit deficiències d’alguns d’aquests elements en pacients que tenen com a base del seu tractament les dietes restrictives. De tots els elements, el Se semblava un candidat ideal, ja que prèviament s’havien descrit deficiències en algunes malalties, com la fenilcetonúria.La importància del Se radica en que és un element essencial per al correcte funcionament de les selenoproteïnes. Per tant, ens semblava d’interès monitoritzar les concentracions de Se tant en sang (com a marcador de dèficit nutricional) com en LCR (com a marcador potencial de mecanismes fisiopatològics en malalties d’àmbit neuropediàtric). Les fites més importants del nostre treball han estat tres: 1. L’estandardització del procediment en sang ha permès establir la tècnica de forma rutinària per al control de trastorns nutricionals al nostre centre. Des d’un punt de vista científic, la conclusió més important és que el dèficit de seleni és comú a les malalties metabòliques amb tractament dietètic i que per tant el seu estudi i suplementació són necessàries per a l’òptim control de les malalties. 2. Respecte a l’estudi del seleni a nivell central, les aportacions han estat més rellevants ja que era un camp poc estudiat. Hem establert per primera vegada valors normals de Se en LCR per a una població pediàtrica i confirmat que el cervell es un òrgan que regula mot bé l’homeòstasi d’aquest element traça, ja que els valors eren independents dels observats en sang. 3. Especialment rellevant creiem que és el perfil que ens va permetre valorar la funció del plexe coroïdal en la síndrome de KSS i el diagnòstic directe d’un cas. Aquest resultats no tan sols tenen un valor per a la pràctica diagnòstica, sinó que obren la porta a la investigació d’altres malalties en les que s’afecta el plexe coroïdal. / PhD STUDENT NAME Mireia Tondo Colomer TESIS TITLE Blood and cerebrospinal fluid selenium status in paediatric patients affected with neurological diseases. SUMMARY Our objective was to study blood and CSF Se status in patients with IEM with dietary treatment and other neurological disorders with different clinical aetiology. In this first study our objective was the establishment of reference values for some trace elements in whole blood. At that time, studies of several trace element status for a paediatric population with different inborn errors of metabolism were scarce. To achieve this aim, we used an Induced Couple Plasma Mass Spectrometer (ICP-MS), which allowed multiple analyses of different trace elements using a small sample volume. Leftover samples form patients who came to our hospital for minor surgery were used as reference population. In this second study our objective was the establishment of cerebrospinal fluid selenium reference values. At that time, no previous studies of Se central nervous system status for a paediatric population existed in the scientific literature. CSF Se concentrations had to be considered in order to find out any possible association between Se status and some neurological functions. Moreover, the implications of Se in the synthesis and function of more than 20 selenoproteins indicated that central nervous system Se status was important. After the establishment of CSF Se reference values for a paediatric population, the next step was to evaluate Se CSF concentrations in paediatric patients with different neurological disorders. Previous studies demonstrated the tight regulation of Se homeostasis in brain as well as the evidence of the role of Se in CNS disorder pathophysiology. In spite of growing evidence of the role of Se in CNS, no studies of Se CNS status in a large cohort of paediatric patients affected with different neurological disorders had been carried out. Finally, our last study derived from the one before which allowed us to identify a group of patients with clearly high CSF Se values. The main common feature was that all patients had a mitochondrial DNA deletion syndrome (Kearns Sayre Syndrome). This group of patients was not included in the previous work and was treated separately.

Multi Relational Data Mining has become popular due to the limitations of propositional problem definition in structured domains and the tendency of storing data in relational databases. However, as patterns involve multiple relations, the search space of possible hypothesis becomes intractably complex. In order to cope with this problem, several relational knowledge discovery systems have been developed employing various search strategies, heuristics and language pattern limitations. In this thesis, Inductive Logic Programming (ILP) based concept discovery is studied and two systems based on a hybrid methodology employing ILP and APRIORI, namely Confidence-based Concept Discovery and Concept Rule Induction System, are proposed. In Confidence-based Concept Discovery and Concept Rule Induction System, the main aim is to relax the strong declarative biases and user-defined specifications. Moreover, this new method directly works on relational databases. In addition to this, the traditional definition of confidence from relational database perspective is modified to express Closed World Assumption in first-order logic. A new confidence-based pruning method based on the improved definition is applied in the APRIORI lattice. Moreover, a new hypothesis evaluation criterion is used for expressing the quality of patterns in the search space. In addition to this, in Concept Rule Induction System, the constructed rule quality is further improved by using an improved generalization metod. Finally, a set of experiments are conducted on real-world problems to evaluate the performance of the proposed method with similar systems in terms of support and confidence.

Many digital signal processing and real-time streaming systems are modeled using dataflow graphs, such as Synchronous Dataflow (SDF) and Cyclo-static Dataflow (CSDF) graphs that allow static analysis and optimization techniques. However, mapping of such descriptions into tightly constrained real-time implementations requires optimization of resource sharing, buffering and scheduling across a multi-dimensional latency-throughput-area objective space. This requires techniques that can find the Pareto-optimal set of implementations for the designer to choose from. In this work, we address the problem of multi-objective mapping and scheduling of SDF and CSDF graphs onto heterogeneous multi-processor platforms. Building on previous work, this thesis extends existing two-stage hybrid heuristics that combine an evolutionary algorithm with an integer linear programming (ILP) model to jointly optimize throughput, area and latency for SDF graphs. The primary contributions of this work include: (1) extension of the ILP model to support CSDFGs with additional buffer size optimizations; (2) a further optimization in the ILP-based scheduling model to achieve a runtime speedup of almost a factor of 10 compared to the existing SDFG formulation; (3) a list scheduling heuristic that replaces the ILP model in the hybrid heuristic to generate Pareto-optimal solutions at significantly decreased runtime while maintaining near-optimality of the solutions within an acceptable gap of 10% when compared to its ILP counterparts. The list scheduling heuristic presented in this work is based on existing modulo scheduling approaches for software pipelining in the compiler domain, but has been extended by introducing a new concept of mobility-based rescheduling before resorting to backtracking. It has been proved in this work that if mobility-based rescheduling is performed, the number of required backtrackings and hence overall complexity and runtime is less. / text

Embedded system designers face multiple challenges in fulfilling the runtime requirements of programs. Effective scheduling of programs is required to extract as much parallelism as possible. These scheduling algorithms must also improve speedup after instruction-set extensions have occurred. Scheduling of dynamic code at run time is made more difficult when the static components of the program are scheduled inefficiently. This research aims to optimize a program’s static code at compile time. This is achieved with four algorithms designed to schedule code at the task and instruction level. Additionally, the algorithms improve scheduling using instruction set extended code on symmetrical homogeneous multiprocessor systems. Using these algorithms, we achieve speedups up to 3.86X over sequential execution for a 4-issue 2-processor system, and show better performance than recent heuristic techniques for small programs. Finally, the algorithms generate speedup values for a 64-point FFT that are similar to the test runs.

Automatic movie analysis is the task of employing Machine Learning methods to the field of screenplays, movie scripts, and motion pictures to facilitate or enable various tasks throughout the entirety of a movie’s life-cycle. From helping with making informed decisions about a new movie script with respect to aspects such as its originality, similarity to other movies, or even commercial viability, all the way to offering consumers new and interesting ways of viewing the final movie, many stages in the life-cycle of a movie stand to benefit from Machine Learning techniques that promise to reduce human effort, time, or both. Within this field of automatic movie analysis, this thesis addresses the task of summarising the content of screenplays, enabling users at any stage to gain a broad understanding of a movie from greatly reduced data. The contributions of this thesis are four-fold: (i)We introduce ScriptBase, a new large-scale data set of original movie scripts, annotated with additional meta-information such as genre and plot tags, cast information, and log- and tag-lines. To our knowledge, Script- Base is the largest data set of its kind, containing scripts and information for almost 1,000 Hollywood movies. (ii) We present a dynamic summarisation model for the screenplay domain, which allows for extraction of highly informative and important scenes from movie scripts. The extracted summaries allow for the content of the original script to stay largely intact and provide the user with its important parts, while greatly reducing the script-reading time. (iii) We extend our summarisation model to capture additional modalities beyond the screenplay text. The model is rendered multi-modal by introducing visual information obtained from the actual movie and by extracting scenes from the movie, allowing users to generate visual summaries of motion pictures. (iv) We devise a novel end-to-end neural network model for generating natural language screenplay overviews. This model enables the user to generate short descriptive and informative texts that capture certain aspects of a movie script, such as its genres, approximate content, or style, allowing them to gain a fast, high-level understanding of the screenplay. Multiple automatic and human evaluations were carried out to assess the performance of our models, demonstrating that they are well-suited for the tasks set out in this thesis, outperforming strong baselines. Furthermore, the ScriptBase data set has started to gain traction, and is currently used by a number of other researchers in the field to tackle various tasks relating to screenplays and their analysis.