Lung Pattern Classification for Interstitial Lung Diseases Using a Deep Convolutional Neural Network

Automated tissue characterization is one of the most crucial components of a computer aided diagnosis (CAD) system for interstitial lung diseases (ILDs). Although much research has been conducted in this field, the problem remains challenging. Deep learning techniques have recently achieved impressive results in a variety of computer vision problems, raising expectations that they might be applied in other domains, such as medical image analysis. In this paper, we propose and evaluate a convolutional neural network (CNN), designed for the classification of ILD patterns. The proposed network consists of 5 convolutional layers with 2×2 kernels and LeakyReLU activations, followed by average pooling with size equal to the size of the final feature maps and three dense layers. The last dense layer has 7 outputs, equivalent to the classes considered: healthy, ground glass opacity (GGO), micronodules, consolidation, reticulation, honeycombing and a combination of GGO/reticulation. To train and evaluate the CNN, we used a dataset of 14696 image patches, derived by 120 CT scans from different scanners and hospitals. To the best of our knowledge, this is the first deep CNN designed for the specific problem. A comparative analysis proved the effectiveness of the proposed CNN against previous methods in a challenging dataset. The classification performance (~85.5%) demonstrated the potential of CNNs in analyzing lung patterns. Future work includes, extending the CNN to three-dimensional data provided by CT volume scans and integrating the proposed method into a CAD system that aims to provide differential diagnosis for ILDs as a supportive tool for radiologists.

The Effect of Proximity, Explicitness, and Representation of Basic Science Information on Student Clinical Problem-Solving

Problem: Medical and veterinary students memorize facts but then have difficulty applying those facts in clinical problem solving. Cognitive engineering research suggests that the inability of medical and veterinary students to infer concepts from facts may be due in part to specific features of how information is represented and organized in educational materials. First, physical separation of pieces of information may increase the cognitive load on the student. Second, information that is necessary but not explicitly stated may also contribute to the student’s cognitive load. Finally, the types of representations – textual or graphical – may also support or hinder the student’s learning process. This may explain why students have difficulty applying biomedical facts in clinical problem solving. Purpose: To test the hypothesis that three specific aspects of expository text – the patial distance between the facts needed to infer a rule, the explicitness of information, and the format of representation – affected the ability of students to solve clinical problems. Setting: The study was conducted in the parasitology laboratory of a college of veterinary medicine in Texas. Sample: The study subjects were a convenience sample consisting of 132 second-year veterinary students who matriculated in 2007. The age of this class upon admission ranged from 20-52, and the gender makeup of this class consisted of approximately 75% females and 25% males. Results: No statistically significant difference in student ability to solve clinical problems was found when relevant facts were placed in proximity, nor when an explicit rule was stated. Further, no statistically significant difference in student ability to solve clinical problems was found when students were given different representations of material, including tables and concept maps. Findings: The findings from this study indicate that the three properties investigated – proximity, explicitness, and representation – had no statistically significant effect on student learning as it relates to clinical problem-solving ability. However, ad hoc observations as well as findings from other researchers suggest that the subjects were probably using rote learning techniques such as memorization, and therefore were not attempting to infer relationships from the factual material in the interventions, unless they were specifically prompted to look for patterns. A serendipitous finding unrelated to the study hypothesis was that those subjects who correctly answered questions regarding functional (non-morphologic) properties, such as mode of transmission and intermediate host, at the family taxonomic level were significantly more likely to correctly answer clinical case scenarios than were subjects who did not correctly answer questions regarding functional properties. These findings suggest a strong relationship (p < .001) between well-organized knowledge of taxonomic functional properties and clinical problem solving ability. Recommendations: Further study should be undertaken investigating the relationship between knowledge of functional taxonomic properties and clinical problem solving ability. In addition, the effect of prompting students to look for patterns in instructional material, followed by the effect of factors that affect cognitive load such as proximity, explicitness, and representation, should be explored.

Methods and Tools for Objective Assessment of Psychomotor Skills in Laparoscopic Surgery

Training and assessment paradigms for laparoscopic surgical skills are evolving from traditional mentor–trainee tutorship towards structured, more objective and safer programs. Accreditation of surgeons requires reaching a consensus on metrics and tasks used to assess surgeons’ psychomotor skills. Ongoing development of tracking systems and software solutions has allowed for the expansion of novel training and assessment means in laparoscopy. The current challenge is to adapt and include these systems within training programs, and to exploit their possibilities for evaluation purposes. This paper describes the state of the art in research on measuring and assessing psychomotor laparoscopic skills. It gives an overview on tracking systems as well as on metrics and advanced statistical and machine learning techniques employed for evaluation purposes. The later ones have a potential to be used as an aid in deciding on the surgical competence level, which is an important aspect when accreditation of the surgeons in particular, and patient safety in general, are considered. The prospective of these methods and tools make them complementary means for surgical assessment of motor skills, especially in the early stages of training. Successful examples such as the Fundamentals of Laparoscopic Surgery should help drive a paradigm change to structured curricula based on objective parameters. These may improve the accreditation of new surgeons, as well as optimize their already overloaded training schedules.

Stress Detection by means of Stress Physiological Template

This paper describes a stress detection system based on fuzzy logic and two physiological signals: Galvanic Skin Response and Heart Rate. Instead of providing a global stress classification, this approach creates an individual stress templates, gathering the behaviour of individuals under situations with different degrees of stress. The proposed method is able to detect stress properly with a rate of 99.5%, being evaluated with a database of 80 individuals. This result improves former approaches in the literature and well-known machine learning techniques like SVM, k-NN, GMM and Linear Discriminant Analysis. Finally, the proposed method is highly suitable for real-time applications

Using Cross-Lingual Explicit Semantic Analysis for Improving Ontology Translation

Semantic Web aims to allow machines to make inferences using the explicit conceptualisations contained in ontologies. By pointing to ontologies, Semantic Web-based applications are able to inter-operate and share common information easily. Nevertheless, multilingual semantic applications are still rare, owing to the fact that most online ontologies are monolingual in English. In order to solve this issue, techniques for ontology localisation and translation are needed. However, traditional machine translation is difficult to apply to ontologies, owing to the fact that ontology labels tend to be quite short in length and linguistically different from the free text paradigm. In this paper, we propose an approach to enhance machine translation of ontologies based on exploiting the well-structured concept descriptions contained in the ontology. In particular, our approach leverages the semantics contained in the ontology by using Cross Lingual Explicit Semantic Analysis (CLESA) for context-based disambiguation in phrase-based Statistical Machine Translation (SMT). The presented work is novel in the sense that application of CLESA in SMT has not been performed earlier to the best of our knowledge.

Semi-supervised subspace clustering and applications to neuroscience

Machine learning techniques are used for extracting valuable knowledge from data. Nowa¬days, these techniques are becoming even more important due to the evolution in data ac¬quisition and storage, which is leading to data with different characteristics that must be exploited. Therefore, advances in data collection must be accompanied with advances in machine learning techniques to solve new challenges that might arise, on both academic and real applications. There are several machine learning techniques depending on both data characteristics and purpose. Unsupervised classification or clustering is one of the most known techniques when data lack of supervision (unlabeled data) and the aim is to discover data groups (clusters) according to their similarity. On the other hand, supervised classification needs data with supervision (labeled data) and its aim is to make predictions about labels of new data. The presence of data labels is a very important characteristic that guides not only the learning task but also other related tasks such as validation. When only some of the available data are labeled whereas the others remain unlabeled (partially labeled data), neither clustering nor supervised classification can be used. This scenario, which is becoming common nowadays because of labeling process ignorance or cost, is tackled with semi-supervised learning techniques. This thesis focuses on the branch of semi-supervised learning closest to clustering, i.e., to discover clusters using available labels as support to guide and improve the clustering process. Another important data characteristic, different from the presence of data labels, is the relevance or not of data features. Data are characterized by features, but it is possible that not all of them are relevant, or equally relevant, for the learning process. A recent clustering tendency, related to data relevance and called subspace clustering, claims that different clusters might be described by different feature subsets. This differs from traditional solutions to data relevance problem, where a single feature subset (usually the complete set of original features) is found and used to perform the clustering process. The proximity of this work to clustering leads to the first goal of this thesis. As commented above, clustering validation is a difficult task due to the absence of data labels. Although there are many indices that can be used to assess the quality of clustering solutions, these validations depend on clustering algorithms and data characteristics. Hence, in the first goal three known clustering algorithms are used to cluster data with outliers and noise, to critically study how some of the most known validation indices behave. The main goal of this work is however to combine semi-supervised clustering with subspace clustering to obtain clustering solutions that can be correctly validated by using either known indices or expert opinions. Two different algorithms are proposed from different points of view to discover clusters characterized by different subspaces. For the first algorithm, available data labels are used for searching for subspaces firstly, before searching for clusters. This algorithm assigns each instance to only one cluster (hard clustering) and is based on mapping known labels to subspaces using supervised classification techniques. Subspaces are then used to find clusters using traditional clustering techniques. The second algorithm uses available data labels to search for subspaces and clusters at the same time in an iterative process. This algorithm assigns each instance to each cluster based on a membership probability (soft clustering) and is based on integrating known labels and the search for subspaces into a model-based clustering approach. The different proposals are tested using different real and synthetic databases, and comparisons to other methods are also included when appropriate. Finally, as an example of real and current application, different machine learning tech¬niques, including one of the proposals of this work (the most sophisticated one) are applied to a task of one of the most challenging biological problems nowadays, the human brain model¬ing. Specifically, expert neuroscientists do not agree with a neuron classification for the brain cortex, which makes impossible not only any modeling attempt but also the day-to-day work without a common way to name neurons. Therefore, machine learning techniques may help to get an accepted solution to this problem, which can be an important milestone for future research in neuroscience. Resumen Las técnicas de aprendizaje automático se usan para extraer información valiosa de datos. Hoy en día, la importancia de estas técnicas está siendo incluso mayor, debido a que la evolución en la adquisición y almacenamiento de datos está llevando a datos con diferentes características que deben ser explotadas. Por lo tanto, los avances en la recolección de datos deben ir ligados a avances en las técnicas de aprendizaje automático para resolver nuevos retos que pueden aparecer, tanto en aplicaciones académicas como reales. Existen varias técnicas de aprendizaje automático dependiendo de las características de los datos y del propósito. La clasificación no supervisada o clustering es una de las técnicas más conocidas cuando los datos carecen de supervisión (datos sin etiqueta), siendo el objetivo descubrir nuevos grupos (agrupaciones) dependiendo de la similitud de los datos. Por otra parte, la clasificación supervisada necesita datos con supervisión (datos etiquetados) y su objetivo es realizar predicciones sobre las etiquetas de nuevos datos. La presencia de las etiquetas es una característica muy importante que guía no solo el aprendizaje sino también otras tareas relacionadas como la validación. Cuando solo algunos de los datos disponibles están etiquetados, mientras que el resto permanece sin etiqueta (datos parcialmente etiquetados), ni el clustering ni la clasificación supervisada se pueden utilizar. Este escenario, que está llegando a ser común hoy en día debido a la ignorancia o el coste del proceso de etiquetado, es abordado utilizando técnicas de aprendizaje semi-supervisadas. Esta tesis trata la rama del aprendizaje semi-supervisado más cercana al clustering, es decir, descubrir agrupaciones utilizando las etiquetas disponibles como apoyo para guiar y mejorar el proceso de clustering. Otra característica importante de los datos, distinta de la presencia de etiquetas, es la relevancia o no de los atributos de los datos. Los datos se caracterizan por atributos, pero es posible que no todos ellos sean relevantes, o igualmente relevantes, para el proceso de aprendizaje. Una tendencia reciente en clustering, relacionada con la relevancia de los datos y llamada clustering en subespacios, afirma que agrupaciones diferentes pueden estar descritas por subconjuntos de atributos diferentes. Esto difiere de las soluciones tradicionales para el problema de la relevancia de los datos, en las que se busca un único subconjunto de atributos (normalmente el conjunto original de atributos) y se utiliza para realizar el proceso de clustering. La cercanía de este trabajo con el clustering lleva al primer objetivo de la tesis. Como se ha comentado previamente, la validación en clustering es una tarea difícil debido a la ausencia de etiquetas. Aunque existen muchos índices que pueden usarse para evaluar la calidad de las soluciones de clustering, estas validaciones dependen de los algoritmos de clustering utilizados y de las características de los datos. Por lo tanto, en el primer objetivo tres conocidos algoritmos se usan para agrupar datos con valores atípicos y ruido para estudiar de forma crítica cómo se comportan algunos de los índices de validación más conocidos. El objetivo principal de este trabajo sin embargo es combinar clustering semi-supervisado con clustering en subespacios para obtener soluciones de clustering que puedan ser validadas de forma correcta utilizando índices conocidos u opiniones expertas. Se proponen dos algoritmos desde dos puntos de vista diferentes para descubrir agrupaciones caracterizadas por diferentes subespacios. Para el primer algoritmo, las etiquetas disponibles se usan para bus¬car en primer lugar los subespacios antes de buscar las agrupaciones. Este algoritmo asigna cada instancia a un único cluster (hard clustering) y se basa en mapear las etiquetas cono-cidas a subespacios utilizando técnicas de clasificación supervisada. El segundo algoritmo utiliza las etiquetas disponibles para buscar de forma simultánea los subespacios y las agru¬paciones en un proceso iterativo. Este algoritmo asigna cada instancia a cada cluster con una probabilidad de pertenencia (soft clustering) y se basa en integrar las etiquetas conocidas y la búsqueda en subespacios dentro de clustering basado en modelos. Las propuestas son probadas utilizando diferentes bases de datos reales y sintéticas, incluyendo comparaciones con otros métodos cuando resulten apropiadas. Finalmente, a modo de ejemplo de una aplicación real y actual, se aplican diferentes técnicas de aprendizaje automático, incluyendo una de las propuestas de este trabajo (la más sofisticada) a una tarea de uno de los problemas biológicos más desafiantes hoy en día, el modelado del cerebro humano. Específicamente, expertos neurocientíficos no se ponen de acuerdo en una clasificación de neuronas para la corteza cerebral, lo que imposibilita no sólo cualquier intento de modelado sino también el trabajo del día a día al no tener una forma estándar de llamar a las neuronas. Por lo tanto, las técnicas de aprendizaje automático pueden ayudar a conseguir una solución aceptada para este problema, lo cual puede ser un importante hito para investigaciones futuras en neurociencia.

An ecological approach to anomaly detection: the EIA Model.

The presented work proposes a new approach for anomaly detection. This approach is based on changes in a population of evolving agents under stress. If conditions are appropriate, changes in the population (modeled by the bioindicators) are representative of the alterations to the environment. This approach, based on an ecological view, improves functionally traditional approaches to the detection of anomalies. To verify this assertion, experiments based on Network Intrussion Detection Systems are presented. The results are compared with the behaviour of other bioinspired approaches and machine learning techniques.

Mars: a personalised mobile activity recognition system

Mobile activity recognition focuses on inferring the current activities of a mobile user by leveraging the sensory data that is available on today’s smart phones. The state of the art in mobile activity recognition uses traditional classification learning techniques. Thus, the learning process typically involves: i) collection of labelled sensory data that is transferred and collated in a centralised repository; ii) model building where the classification model is trained and tested using the collected data; iii) a model deployment stage where the learnt model is deployed on-board a mobile device for identifying activities based on new sensory data. In this paper, we demonstrate the Mobile Activity Recognition System (MARS) where for the first time the model is built and continuously updated on-board the mobile device itself using data stream mining. The advantages of the on-board approach are that it allows model personalisation and increased privacy as the data is not sent to any external site. Furthermore, when the user or its activity profile changes MARS enables promptly adaptation. MARS has been implemented on the Android platform to demonstrate that it can achieve accurate mobile activity recognition. Moreover, we can show in practise that MARS quickly adapts to user profile changes while at the same time being scalable and efficient in terms of consumption of the device resources.

Sparse spectral analysis of atrial fibrillation electrograms

Atrial fibrillation (AF) is a common heart disorder. One of the most prominent hypothesis about its initiation and maintenance considers multiple uncoordinated activation foci inside the atrium. However, the implicit assumption behind all the signal processing techniques used for AF, such as dominant frequency and organization analysis, is the existence of a single regular component in the observed signals. In this paper we take into account the existence of multiple foci, performing a spectral analysis to detect their number and frequencies. In order to obtain a cleaner signal on which the spectral analysis can be performed, we introduce sparsity-aware learning techniques to infer the spike trains corresponding to the activations. The good performance of the proposed algorithm is demonstrated both on synthetic and real data. RESUMEN. Algoritmo basado en técnicas de regresión dispersa para la extracción de las señales cardiacas en pacientes con fibrilación atrial (AF).

Development and implementation of a virtual laboratory for training process improvement in the mechanics of continuous media

The project arises from the need to develop improved teaching methodologies in field of the mechanics of continuous media. The objective is to offer the student a learning process to acquire the necessary theoretical knowledge, cognitive skills and the responsibility and autonomy to professional development in this area. Traditionally the teaching of the concepts of these subjects was performed through lectures and laboratory practice. During these lessons the students attitude was usually passive, and therefore their effectiveness was poor. The proposed methodology has already been successfully employed in universities like University Bochum, Germany, University the South Australia and aims to improve the effectiveness of knowledge acquisition through use by the student of a virtual laboratory. This laboratory allows to adapt the curricula and learning techniques to the European Higher Education and improve current learning processes in the University School of Public Works Engineers -EUITOP- of the Technical University of Madrid -UPM-, due there are not laboratories in this specialization. The virtual space is created using a software platform built on OpenSim, manages 3D virtual worlds, and, language LSL -Linden Scripting Language-, which imprints specific powers to objects. The student or user can access this virtual world through their avatar -your character in the virtual world- and can perform practices within the space created for the purpose, at any time, just with computer with internet access and viewfinder. The virtual laboratory has three partitions. The virtual meeting rooms, where the avatar can interact with peers, solve problems and exchange existing documentation in the virtual library. The interactive game room, where the avatar is has to resolve a number of issues in time. And the video room where students can watch instructional videos and receive group lessons. Each audiovisual interactive element is accompanied by explanations framing it within the area of knowledge and enables students to begin to acquire a vocabulary and practice of the profession for which they are being formed. Plane elasticity concepts are introduced from the tension and compression testing of test pieces of steel and concrete. The behavior of reticulated and articulated structures is reinforced by some interactive games and concepts of tension, compression, local and global buckling will by tests to break articulated structures. Pure bending concepts, simple and composite torsion will be studied by observing a flexible specimen. Earthquake resistant design of buildings will be checked by a laboratory test video.

Plataforma de evaluación de algoritmos para la identificación de usuarios

La minería de datos es un campo de las ciencias de la computación referido al proceso que intenta descubrir patrones en grandes volúmenes de datos. La minería de datos busca generar información similar a la que podría producir un experto humano. Además es el proceso de descubrir conocimientos interesantes, como patrones, asociaciones, cambios, anomalías y estructuras significativas a partir de grandes cantidades de datos almacenadas en bases de datos, data warehouses o cualquier otro medio de almacenamiento de información. El aprendizaje automático o aprendizaje de máquinas es una rama de la Inteligencia artificial cuyo objetivo es desarrollar técnicas que permitan a las computadoras aprender. De forma más concreta, se trata de crear programas capaces de generalizar comportamientos a partir de una información no estructurada suministrada en forma de ejemplos. La minería de datos utiliza métodos de aprendizaje automático para descubrir y enumerar patrones presentes en los datos. En los últimos años se han aplicado las técnicas de clasificación y aprendizaje automático en un número elevado de ámbitos como el sanitario, comercial o de seguridad. Un ejemplo muy actual es la detección de comportamientos y transacciones fraudulentas en bancos. Una aplicación de interés es el uso de las técnicas desarrolladas para la detección de comportamientos fraudulentos en la identificación de usuarios existentes en el interior de entornos inteligentes sin necesidad de realizar un proceso de autenticación. Para comprobar que estas técnicas son efectivas durante la fase de análisis de una determinada solución, es necesario crear una plataforma que de soporte al desarrollo, validación y evaluación de algoritmos de aprendizaje y clasificación en los entornos de aplicación bajo estudio. El proyecto planteado está definido para la creación de una plataforma que permita evaluar algoritmos de aprendizaje automático como mecanismos de identificación en espacios inteligentes. Se estudiarán tanto los algoritmos propios de este tipo de técnicas como las plataformas actuales existentes para definir un conjunto de requisitos específicos de la plataforma a desarrollar. Tras el análisis se desarrollará parcialmente la plataforma. Tras el desarrollo se validará con pruebas de concepto y finalmente se verificará en un entorno de investigación a definir. ABSTRACT. The data mining is a field of the sciences of the computation referred to the process that it tries to discover patterns in big volumes of information. The data mining seeks to generate information similar to the one that a human expert might produce. In addition it is the process of discovering interesting knowledge, as patterns, associations, changes, abnormalities and significant structures from big quantities of information stored in databases, data warehouses or any other way of storage of information. The machine learning is a branch of the artificial Intelligence which aim is to develop technologies that they allow the computers to learn. More specifically, it is a question of creating programs capable of generalizing behaviors from not structured information supplied in the form of examples. The data mining uses methods of machine learning to discover and to enumerate present patterns in the information. In the last years there have been applied classification and machine learning techniques in a high number of areas such as healthcare, commercial or security. A very current example is the detection of behaviors and fraudulent transactions in banks. An application of interest is the use of the techniques developed for the detection of fraudulent behaviors in the identification of existing Users inside intelligent environments without need to realize a process of authentication. To verify these techniques are effective during the phase of analysis of a certain solution, it is necessary to create a platform that support the development, validation and evaluation of algorithms of learning and classification in the environments of application under study. The project proposed is defined for the creation of a platform that allows evaluating algorithms of machine learning as mechanisms of identification in intelligent spaces. There will be studied both the own algorithms of this type of technologies and the current existing platforms to define a set of specific requirements of the platform to develop. After the analysis the platform will develop partially. After the development it will be validated by prove of concept and finally verified in an environment of investigation that would be define.

A telecom analytics framework for dynamic quality of service management

Since the beginning of Internet, Internet Service Providers (ISP) have seen the need of giving to users? traffic different treatments defined by agree- ments between ISP and customers. This procedure, known as Quality of Service Management, has not much changed in the last years (DiffServ and Deep Pack-et Inspection have been the most chosen mechanisms). However, the incremen-tal growth of Internet users and services jointly with the application of recent Ma- chine Learning techniques, open up the possibility of going one step for-ward in the smart management of network traffic. In this paper, we first make a survey of current tools and techniques for QoS Management. Then we intro-duce clustering and classifying Machine Learning techniques for traffic charac-terization and the concept of Quality of Experience. Finally, with all these com-ponents, we present a brand new framework that will manage in a smart way Quality of Service in a telecom Big Data based scenario, both for mobile and fixed communications.

Application of PZT and FBG impedance sensors for structural health monitoring of reinforced concrete beams strengthened with FRP composite materials

Esta Tesis tiene como objetivo principal el desarrollo de métodos de identificación del daño que sean robustos y fiables, enfocados a sistemas estructurales experimentales, fundamentalmente a las estructuras de hormigón armado reforzadas externamente con bandas fibras de polímeros reforzados (FRP). El modo de fallo de este tipo de sistema estructural es crítico, pues generalmente es debido a un despegue repentino y frágil de la banda del refuerzo FRP originado en grietas intermedias causadas por la flexión. La detección de este despegue en su fase inicial es fundamental para prevenir fallos futuros, que pueden ser catastróficos. Inicialmente, se lleva a cabo una revisión del método de la Impedancia Electro-Mecánica (EMI), de cara a exponer sus capacidades para la detección de daño. Una vez la tecnología apropiada es seleccionada, lo que incluye un analizador de impedancias así como novedosos sensores PZT para monitorización inteligente, se ha diseñado un procedimiento automático basado en los registros de impedancias de distintas estructuras de laboratorio. Basándonos en el hecho de que las mediciones de impedancias son posibles gracias a una colocación adecuada de una red de sensores PZT, la estimación de la presencia de daño se realiza analizando los resultados de distintos indicadores de daño obtenidos de la literatura. Para que este proceso sea automático y que no sean necesarios conocimientos previos sobre el método EMI para realizar un experimento, se ha diseñado e implementado un Interfaz Gráfico de Usuario, transformando la medición de impedancias en un proceso fácil e intuitivo. Se evalúa entonces el daño a través de los correspondientes índices de daño, intentando estimar no sólo su severidad, sino también su localización aproximada. El desarrollo de estos experimentos en cualquier estructura genera grandes cantidades de datos que han de ser procesados, y algunas veces los índices de daño no son suficientes para una evaluación completa de la integridad de una estructura. En la mayoría de los casos se pueden encontrar patrones de daño en los datos, pero no se tiene información a priori del estado de la estructura. En este punto, se ha hecho una importante investigación en técnicas de reconocimiento de patrones particularmente en aprendizaje no supervisado, encontrando aplicaciones interesantes en el campo de la medicina. De ahí surge una idea creativa e innovadora: detectar y seguir la evolución del daño en distintas estructuras como si se tratase de un cáncer propagándose por el cuerpo humano. En ese sentido, las lecturas de impedancias se emplean como información intrínseca de la salud de la propia estructura, de forma que se pueden aplicar las mismas técnicas que las empleadas en la investigación del cáncer. En este caso, se ha aplicado un algoritmo de clasificación jerárquica dado que ilustra además la clasificación de los datos de forma gráfica, incluyendo información cualitativa y cuantitativa sobre el daño. Se ha investigado la efectividad de este procedimiento a través de tres estructuras de laboratorio, como son una viga de aluminio, una unión atornillada de aluminio y un bloque de hormigón reforzado con FRP. La primera ayuda a mostrar la efectividad del método en sencillos escenarios de daño simple y múltiple, de forma que las conclusiones extraídas se aplican sobre los otros dos, diseñados para simular condiciones de despegue en distintas estructuras. Demostrada la efectividad del método de clasificación jerárquica de lecturas de impedancias, se aplica el procedimiento sobre las estructuras de hormigón armado reforzadas con bandas de FRP objeto de esta tesis, detectando y clasificando cada estado de daño. Finalmente, y como alternativa al anterior procedimiento, se propone un método para la monitorización continua de la interfase FRP-Hormigón, a través de una red de sensores FBG permanentemente instalados en dicha interfase. De esta forma, se obtienen medidas de deformación de la interfase en condiciones de carga continua, para ser implementadas en un modelo de optimización multiobjetivo, cuya solución se haya por medio de una expansión multiobjetivo del método Particle Swarm Optimization (PSO). La fiabilidad de este último método de detección se investiga a través de sendos ejemplos tanto numéricos como experimentales. ABSTRACT This thesis aims to develop robust and reliable damage identification methods focused on experimental structural systems, in particular Reinforced Concrete (RC) structures externally strengthened with Fiber Reinforced Polymers (FRP) strips. The failure mode of this type of structural system is critical, since it is usually due to sudden and brittle debonding of the FRP reinforcement originating from intermediate flexural cracks. Detection of the debonding in its initial stage is essential thus to prevent future failure, which might be catastrophic. Initially, a revision of the Electro-Mechanical Impedance (EMI) method is carried out, in order to expose its capabilities for local damage detection. Once the appropriate technology is selected, which includes impedance analyzer as well as novel PZT sensors for smart monitoring, an automated procedure has been design based on the impedance signatures of several lab-scale structures. On the basis that capturing impedance measurements is possible thanks to an adequately deployed PZT sensor network, the estimation of damage presence is done by analyzing the results of different damage indices obtained from the literature. In order to make this process automatic so that it is not necessary a priori knowledge of the EMI method to carry out an experimental test, a Graphical User Interface has been designed, turning the impedance measurements into an easy and intuitive procedure. Damage is then assessed through the analysis of the corresponding damage indices, trying to estimate not only the damage severity, but also its approximate location. The development of these tests on any kind of structure generates large amounts of data to be processed, and sometimes the information provided by damage indices is not enough to achieve a complete analysis of the structural health condition. In most of the cases, some damage patterns can be found in the data, but none a priori knowledge of the health condition is given for any structure. At this point, an important research on pattern recognition techniques has been carried out, particularly on unsupervised learning techniques, finding interesting applications in the medicine field. From this investigation, a creative and innovative idea arose: to detect and track the evolution of damage in different structures, as if it were a cancer propagating through a human body. In that sense, the impedance signatures are used to give intrinsic information of the health condition of the structure, so that the same clustering algorithms applied in the cancer research can be applied to the problem addressed in this dissertation. Hierarchical clustering is then applied since it also provides a graphical display of the clustered data, including quantitative and qualitative information about damage. The performance of this approach is firstly investigated using three lab-scale structures, such as a simple aluminium beam, a bolt-jointed aluminium beam and an FRP-strengthened concrete specimen. The first one shows the performance of the method on simple single and multiple damage scenarios, so that the first conclusions can be extracted and applied to the other two experimental tests, which are designed to simulate a debonding condition on different structures. Once the performance of the impedance-based hierarchical clustering method is proven to be successful, it is then applied to the structural system studied in this dissertation, the RC structures externally strengthened with FRP strips, where the debonding failure in the interface between the FRP and the concrete is successfully detected and classified, proving thus the feasibility of this method. Finally, as an alternative to the previous approach, a continuous monitoring procedure of the FRP-Concrete interface is proposed, based on an FBGsensors Network permanently deployed within that interface. In this way, strain measurements can be obtained under controlled loading conditions, and then they are used in order to implement a multi-objective model updating method solved by a multi-objective expansion of the Particle Swarm Optimization (PSO) method. The feasibility of this last proposal is investigated and successfully proven on both numerical and experimental RC beams strengthened with FRP.

Contributions to Speech Analytics based on Speech Recognition and Topic Identification

La última década ha sido testigo de importantes avances en el campo de la tecnología de reconocimiento de voz. Los sistemas comerciales existentes actualmente poseen la capacidad de reconocer habla continua de múltiples locutores, consiguiendo valores aceptables de error, y sin la necesidad de realizar procedimientos explícitos de adaptación. A pesar del buen momento que vive esta tecnología, el reconocimiento de voz dista de ser un problema resuelto. La mayoría de estos sistemas de reconocimiento se ajustan a dominios particulares y su eficacia depende de manera significativa, entre otros muchos aspectos, de la similitud que exista entre el modelo de lenguaje utilizado y la tarea específica para la cual se está empleando. Esta dependencia cobra aún más importancia en aquellos escenarios en los cuales las propiedades estadísticas del lenguaje varían a lo largo del tiempo, como por ejemplo, en dominios de aplicación que involucren habla espontánea y múltiples temáticas. En los últimos años se ha evidenciado un constante esfuerzo por mejorar los sistemas de reconocimiento para tales dominios. Esto se ha hecho, entre otros muchos enfoques, a través de técnicas automáticas de adaptación. Estas técnicas son aplicadas a sistemas ya existentes, dado que exportar el sistema a una nueva tarea o dominio puede requerir tiempo a la vez que resultar costoso. Las técnicas de adaptación requieren fuentes adicionales de información, y en este sentido, el lenguaje hablado puede aportar algunas de ellas. El habla no sólo transmite un mensaje, también transmite información acerca del contexto en el cual se desarrolla la comunicación hablada (e.g. acerca del tema sobre el cual se está hablando). Por tanto, cuando nos comunicamos a través del habla, es posible identificar los elementos del lenguaje que caracterizan el contexto, y al mismo tiempo, rastrear los cambios que ocurren en estos elementos a lo largo del tiempo. Esta información podría ser capturada y aprovechada por medio de técnicas de recuperación de información (information retrieval) y de aprendizaje de máquina (machine learning). Esto podría permitirnos, dentro del desarrollo de mejores sistemas automáticos de reconocimiento de voz, mejorar la adaptación de modelos del lenguaje a las condiciones del contexto, y por tanto, robustecer al sistema de reconocimiento en dominios con condiciones variables (tales como variaciones potenciales en el vocabulario, el estilo y la temática). En este sentido, la principal contribución de esta Tesis es la propuesta y evaluación de un marco de contextualización motivado por el análisis temático y basado en la adaptación dinámica y no supervisada de modelos de lenguaje para el robustecimiento de un sistema automático de reconocimiento de voz. Esta adaptación toma como base distintos enfoque de los sistemas mencionados (de recuperación de información y aprendizaje de máquina) mediante los cuales buscamos identificar las temáticas sobre las cuales se está hablando en una grabación de audio. Dicha identificación, por lo tanto, permite realizar una adaptación del modelo de lenguaje de acuerdo a las condiciones del contexto. El marco de contextualización propuesto se puede dividir en dos sistemas principales: un sistema de identificación de temática y un sistema de adaptación dinámica de modelos de lenguaje. Esta Tesis puede describirse en detalle desde la perspectiva de las contribuciones particulares realizadas en cada uno de los campos que componen el marco propuesto: _ En lo referente al sistema de identificación de temática, nos hemos enfocado en aportar mejoras a las técnicas de pre-procesamiento de documentos, asimismo en contribuir a la definición de criterios más robustos para la selección de index-terms. – La eficiencia de los sistemas basados tanto en técnicas de recuperación de información como en técnicas de aprendizaje de máquina, y específicamente de aquellos sistemas que particularizan en la tarea de identificación de temática, depende, en gran medida, de los mecanismos de preprocesamiento que se aplican a los documentos. Entre las múltiples operaciones que hacen parte de un esquema de preprocesamiento, la selección adecuada de los términos de indexado (index-terms) es crucial para establecer relaciones semánticas y conceptuales entre los términos y los documentos. Este proceso también puede verse afectado, o bien por una mala elección de stopwords, o bien por la falta de precisión en la definición de reglas de lematización. En este sentido, en este trabajo comparamos y evaluamos diferentes criterios para el preprocesamiento de los documentos, así como también distintas estrategias para la selección de los index-terms. Esto nos permite no sólo reducir el tamaño de la estructura de indexación, sino también mejorar el proceso de identificación de temática. – Uno de los aspectos más importantes en cuanto al rendimiento de los sistemas de identificación de temática es la asignación de diferentes pesos a los términos de acuerdo a su contribución al contenido del documento. En este trabajo evaluamos y proponemos enfoques alternativos a los esquemas tradicionales de ponderado de términos (tales como tf-idf ) que nos permitan mejorar la especificidad de los términos, así como también discriminar mejor las temáticas de los documentos. _ Respecto a la adaptación dinámica de modelos de lenguaje, hemos dividimos el proceso de contextualización en varios pasos. – Para la generación de modelos de lenguaje basados en temática, proponemos dos tipos de enfoques: un enfoque supervisado y un enfoque no supervisado. En el primero de ellos nos basamos en las etiquetas de temática que originalmente acompañan a los documentos del corpus que empleamos. A partir de estas, agrupamos los documentos que forman parte de la misma temática y generamos modelos de lenguaje a partir de dichos grupos. Sin embargo, uno de los objetivos que se persigue en esta Tesis es evaluar si el uso de estas etiquetas para la generación de modelos es óptimo en términos del rendimiento del reconocedor. Por esta razón, nosotros proponemos un segundo enfoque, un enfoque no supervisado, en el cual el objetivo es agrupar, automáticamente, los documentos en clusters temáticos, basándonos en la similaridad semántica existente entre los documentos. Por medio de enfoques de agrupamiento conseguimos mejorar la cohesión conceptual y semántica en cada uno de los clusters, lo que a su vez nos permitió refinar los modelos de lenguaje basados en temática y mejorar el rendimiento del sistema de reconocimiento. – Desarrollamos diversas estrategias para generar un modelo de lenguaje dependiente del contexto. Nuestro objetivo es que este modelo refleje el contexto semántico del habla, i.e. las temáticas más relevantes que se están discutiendo. Este modelo es generado por medio de la interpolación lineal entre aquellos modelos de lenguaje basados en temática que estén relacionados con las temáticas más relevantes. La estimación de los pesos de interpolación está basada principalmente en el resultado del proceso de identificación de temática. – Finalmente, proponemos una metodología para la adaptación dinámica de un modelo de lenguaje general. El proceso de adaptación tiene en cuenta no sólo al modelo dependiente del contexto sino también a la información entregada por el proceso de identificación de temática. El esquema usado para la adaptación es una interpolación lineal entre el modelo general y el modelo dependiente de contexto. Estudiamos también diferentes enfoques para determinar los pesos de interpolación entre ambos modelos. Una vez definida la base teórica de nuestro marco de contextualización, proponemos su aplicación dentro de un sistema automático de reconocimiento de voz. Para esto, nos enfocamos en dos aspectos: la contextualización de los modelos de lenguaje empleados por el sistema y la incorporación de información semántica en el proceso de adaptación basado en temática. En esta Tesis proponemos un marco experimental basado en una arquitectura de reconocimiento en ‘dos etapas’. En la primera etapa, empleamos sistemas basados en técnicas de recuperación de información y aprendizaje de máquina para identificar las temáticas sobre las cuales se habla en una transcripción de un segmento de audio. Esta transcripción es generada por el sistema de reconocimiento empleando un modelo de lenguaje general. De acuerdo con la relevancia de las temáticas que han sido identificadas, se lleva a cabo la adaptación dinámica del modelo de lenguaje. En la segunda etapa de la arquitectura de reconocimiento, usamos este modelo adaptado para realizar de nuevo el reconocimiento del segmento de audio. Para determinar los beneficios del marco de trabajo propuesto, llevamos a cabo la evaluación de cada uno de los sistemas principales previamente mencionados. Esta evaluación es realizada sobre discursos en el dominio de la política usando la base de datos EPPS (European Parliamentary Plenary Sessions - Sesiones Plenarias del Parlamento Europeo) del proyecto europeo TC-STAR. Analizamos distintas métricas acerca del rendimiento de los sistemas y evaluamos las mejoras propuestas con respecto a los sistemas de referencia. ABSTRACT The last decade has witnessed major advances in speech recognition technology. Today’s commercial systems are able to recognize continuous speech from numerous speakers, with acceptable levels of error and without the need for an explicit adaptation procedure. Despite this progress, speech recognition is far from being a solved problem. Most of these systems are adjusted to a particular domain and their efficacy depends significantly, among many other aspects, on the similarity between the language model used and the task that is being addressed. This dependence is even more important in scenarios where the statistical properties of the language fluctuates throughout the time, for example, in application domains involving spontaneous and multitopic speech. Over the last years there has been an increasing effort in enhancing the speech recognition systems for such domains. This has been done, among other approaches, by means of techniques of automatic adaptation. These techniques are applied to the existing systems, specially since exporting the system to a new task or domain may be both time-consuming and expensive. Adaptation techniques require additional sources of information, and the spoken language could provide some of them. It must be considered that speech not only conveys a message, it also provides information on the context in which the spoken communication takes place (e.g. on the subject on which it is being talked about). Therefore, when we communicate through speech, it could be feasible to identify the elements of the language that characterize the context, and at the same time, to track the changes that occur in those elements over time. This information can be extracted and exploited through techniques of information retrieval and machine learning. This allows us, within the development of more robust speech recognition systems, to enhance the adaptation of language models to the conditions of the context, thus strengthening the recognition system for domains under changing conditions (such as potential variations in vocabulary, style and topic). In this sense, the main contribution of this Thesis is the proposal and evaluation of a framework of topic-motivated contextualization based on the dynamic and non-supervised adaptation of language models for the enhancement of an automatic speech recognition system. This adaptation is based on an combined approach (from the perspective of both information retrieval and machine learning fields) whereby we identify the topics that are being discussed in an audio recording. The topic identification, therefore, enables the system to perform an adaptation of the language model according to the contextual conditions. The proposed framework can be divided in two major systems: a topic identification system and a dynamic language model adaptation system. This Thesis can be outlined from the perspective of the particular contributions made in each of the fields that composes the proposed framework: _ Regarding the topic identification system, we have focused on the enhancement of the document preprocessing techniques in addition to contributing in the definition of more robust criteria for the selection of index-terms. – Within both information retrieval and machine learning based approaches, the efficiency of topic identification systems, depends, to a large extent, on the mechanisms of preprocessing applied to the documents. Among the many operations that encloses the preprocessing procedures, an adequate selection of index-terms is critical to establish conceptual and semantic relationships between terms and documents. This process might also be weakened by a poor choice of stopwords or lack of precision in defining stemming rules. In this regard we compare and evaluate different criteria for preprocessing the documents, as well as for improving the selection of the index-terms. This allows us to not only reduce the size of the indexing structure but also to strengthen the topic identification process. – One of the most crucial aspects, in relation to the performance of topic identification systems, is to assign different weights to different terms depending on their contribution to the content of the document. In this sense we evaluate and propose alternative approaches to traditional weighting schemes (such as tf-idf ) that allow us to improve the specificity of terms, and to better identify the topics that are related to documents. _ Regarding the dynamic language model adaptation, we divide the contextualization process into different steps. – We propose supervised and unsupervised approaches for the generation of topic-based language models. The first of them is intended to generate topic-based language models by grouping the documents, in the training set, according to the original topic labels of the corpus. Nevertheless, a goal of this Thesis is to evaluate whether or not the use of these labels to generate language models is optimal in terms of recognition accuracy. For this reason, we propose a second approach, an unsupervised one, in which the objective is to group the data in the training set into automatic topic clusters based on the semantic similarity between the documents. By means of clustering approaches we expect to obtain a more cohesive association of the documents that are related by similar concepts, thus improving the coverage of the topic-based language models and enhancing the performance of the recognition system. – We develop various strategies in order to create a context-dependent language model. Our aim is that this model reflects the semantic context of the current utterance, i.e. the most relevant topics that are being discussed. This model is generated by means of a linear interpolation between the topic-based language models related to the most relevant topics. The estimation of the interpolation weights is based mainly on the outcome of the topic identification process. – Finally, we propose a methodology for the dynamic adaptation of a background language model. The adaptation process takes into account the context-dependent model as well as the information provided by the topic identification process. The scheme used for the adaptation is a linear interpolation between the background model and the context-dependent one. We also study different approaches to determine the interpolation weights used in this adaptation scheme. Once we defined the basis of our topic-motivated contextualization framework, we propose its application into an automatic speech recognition system. We focus on two aspects: the contextualization of the language models used by the system, and the incorporation of semantic-related information into a topic-based adaptation process. To achieve this, we propose an experimental framework based in ‘a two stages’ recognition architecture. In the first stage of the architecture, Information Retrieval and Machine Learning techniques are used to identify the topics in a transcription of an audio segment. This transcription is generated by the recognition system using a background language model. According to the confidence on the topics that have been identified, the dynamic language model adaptation is carried out. In the second stage of the recognition architecture, an adapted language model is used to re-decode the utterance. To test the benefits of the proposed framework, we carry out the evaluation of each of the major systems aforementioned. The evaluation is conducted on speeches of political domain using the EPPS (European Parliamentary Plenary Sessions) database from the European TC-STAR project. We analyse several performance metrics that allow us to compare the improvements of the proposed systems against the baseline ones.

Automatização do processo de seleção de transformações para otimização do tempo de execução por meio de aprendizado de máquina no arcabouço da LLVM.

A rápida evolução do hardware demanda uma evolução contínua dos compiladores. Um processo de ajuste deve ser realizado pelos projetistas de compiladores para garantir que o código gerado pelo compilador mantenha uma determinada qualidade, seja em termos de tempo de processamento ou outra característica pré-definida. Este trabalho visou automatizar o processo de ajuste de compiladores por meio de técnicas de aprendizado de máquina. Como resultado os planos de compilação obtidos usando aprendizado de máquina com as características propostas produziram código para programas cujos valores para os tempos de execução se aproximaram daqueles seguindo o plano padrão utilizado pela LLVM.