IDT-3D: Identification and tracking in controlled environments using a 3D unified user interface

Identification and tracking of objects in specific environments such as harbors or security areas is a matter of great importance nowadays. With this purpose, numerous systems based on different technologies have been developed, resulting in a great amount of gathered data displayed through a variety of interfaces. Such amount of information has to be evaluated by human operators in order to take the correct decisions, sometimes under highly critical situations demanding both speed and accuracy. In order to face this problem we describe IDT-3D, a platform for identification and tracking of vessels in a harbour environment able to represent fused information in real time using a Virtual Reality application. The effectiveness of using IDT-3D as an integrated surveillance system is currently under evaluation. Preliminary results point to a significant decrease in the times of reaction and decision making of operators facing up a critical situation. Although the current application focus of IDT-3D is quite specific, the results of this research could be extended to the identification and tracking of targets in other controlled environments of interest as coastlines, borders or even urban areas.

DAEDALUS at ImageCLEF 2011 Plant Identification Task: Using SIFT Keypoints for Object Detection

This paper describes the participation of DAEDALUS at ImageCLEF 2011 Plant Identification task. The task is evaluated as a supervised classification problem over 71 tree species from the French Mediterranean area used as class labels, based on visual content from scan, scan-like and natural photo images. Our approach to this task is to build a classifier based on the detection of keypoints from the images extracted using Lowe’s Scale Invariant Feature Transform (SIFT) algorithm. Although our overall classification score is very low as compared to other participant groups, the main conclusion that can be drawn is that SIFT keypoints seem to work significantly better for photos than for the other image types, so our approach may be a feasible strategy for the classification of this kind of visual content.

Computational Methods for Identification of Vibrating Structures

System identification deals with the problem of building mathematical models of dynamical systems based on observed data from the system" [1]. In the context of civil engineering, the system refers to a large scale structure such as a building, bridge, or an offshore structure, and identification mostly involves the determination of modal parameters (the natural frequencies, damping ratios, and mode shapes). This paper presents some modal identification results obtained using a state-of-the-art time domain system identification method (data-driven stochastic subspace algorithms [2]) applied to the output-only data measured in a steel arch bridge. First, a three dimensional finite element model was developed for the numerical analysis of the structure using ANSYS. Modal analysis was carried out and modal parameters were extracted in the frequency range of interest, 0-10 Hz. The results obtained from the finite element modal analysis were used to determine the location of the sensors. After that, ambient vibration tests were conducted during April 23-24, 2009. The response of the structure was measured using eight accelerometers. Two stations of three sensors were formed (triaxial stations). These sensors were held stationary for reference during the test. The two remaining sensors were placed at the different measurement points along the bridge deck, in which only vertical and transversal measurements were conducted (biaxial stations). Point estimate and interval estimate have been carried out in the state space model using these ambient vibration measurements. In the case of parametric models (like state space), the dynamic behaviour of a system is described using mathematical models. Then, mathematical relationships can be established between modal parameters and estimated point parameters (thus, it is common to use experimental modal analysis as a synonym for system identification). Stable modal parameters are found using a stabilization diagram. Furthermore, this paper proposes a method for assessing the precision of estimates of the parameters of state-space models (confidence interval). This approach employs the nonparametric bootstrap procedure [3] and is applied to subspace parameter estimation algorithm. Using bootstrap results, a plot similar to a stabilization diagram is developed. These graphics differentiate system modes from spurious noise modes for a given order system. Additionally, using the modal assurance criterion, the experimental modes obtained have been compared with those evaluated from a finite element analysis. A quite good agreement between numerical and experimental results is observed.

Combining a Similar Coefficient Identification Algorithm with the Boundary Element Method

The boundary element method (BEM) has been applied successfully to many engineering problems during the last decades. Compared with domain type methods like the finite element method (FEM) or the finite difference method (FDM) the BEM can handle problems where the medium extends to infinity much easier than domain type methods as there is no need to develop special boundary conditions (quiet or absorbing boundaries) or infinite elements at the boundaries introduced to limit the domain studied. The determination of the dynamic stiffness of arbitrarily shaped footings is just one of these fields where the BEM has been the method of choice, especially in the 1980s. With the continuous development of computer technology and the available hardware equipment the size of the problems under study grew and, as the flop count for solving the resulting linear system of equations grows with the third power of the number of equations, there was a need for the development of iterative methods with better performance. In [1] the GMRES algorithm was presented which is now widely used for implementations of the collocation BEM. While the FEM results in sparsely populated coefficient matrices, the BEM leads, in general, to fully or densely populated ones, depending on the number of subregions, posing a serious memory problem even for todays computers. If the geometry of the problem permits the surface of the domain to be meshed with equally shaped elements a lot of the resulting coefficients will be calculated and stored repeatedly. The present paper shows how these unnecessary operations can be avoided reducing the calculation time as well as the storage requirement. To this end a similar coefficient identification algorithm (SCIA), has been developed and implemented in a program written in Fortran 90. The vertical dynamic stiffness of a single pile in layered soil has been chosen to test the performance of the implementation. The results obtained with the 3-d model may be compared with those obtained with an axisymmetric formulation which are considered to be the reference values as the mesh quality is much better. The entire 3D model comprises more than 35000 dofs being a soil region with 21168 dofs the biggest single region. Note that the memory necessary to store all coefficients of this single region is about 6.8 GB, an amount which is usually not available with personal computers. In the problem under study the interface zone between the two adjacent soil regions as well as the surface of the top layer may be meshed with equally sized elements. In this case the application of the SCIA leads to an important reduction in memory requirements. The maximum memory used during the calculation has been reduced to 1.2 GB. The application of the SCIA thus permits problems to be solved on personal computers which otherwise would require much more powerful hardware.

Novel fast random search clustering approach for mixing matrix identification in MIMO linear blind inverse problems with sparse inputs

In this paper we propose a novel fast random search clustering (RSC) algorithm for mixing matrix identification in multiple input multiple output (MIMO) linear blind inverse problems with sparse inputs. The proposed approach is based on the clustering of the observations around the directions given by the columns of the mixing matrix that occurs typically for sparse inputs. Exploiting this fact, the RSC algorithm proceeds by parameterizing the mixing matrix using hyperspherical coordinates, randomly selecting candidate basis vectors (i.e. clustering directions) from the observations, and accepting or rejecting them according to a binary hypothesis test based on the Neyman–Pearson criterion. The RSC algorithm is not tailored to any specific distribution for the sources, can deal with an arbitrary number of inputs and outputs (thus solving the difficult under-determined problem), and is applicable to both instantaneous and convolutive mixtures. Extensive simulations for synthetic and real data with different number of inputs and outputs, data size, sparsity factors of the inputs and signal to noise ratios confirm the good performance of the proposed approach under moderate/high signal to noise ratios. RESUMEN. Método de separación ciega de fuentes para señales dispersas basado en la identificación de la matriz de mezcla mediante técnicas de "clustering" aleatorio.

Speed-independent gait identification for mobile devices

Due to the intensive use of mobile phones for diferent purposes, these devices usually contain condential information which must not be accessed by another person apart from the owner of the device. Furthermore, the new generation phones commonly incorporate an accelerometer which may be used to capture the acceleration signals produced as a result of owner s gait. Nowadays, gait identication in basis of acceleration signals is being considered as a new biometric technique which allows blocking the device when another person is carrying it. Although distance based approaches as Euclidean distance or dynamic time warping have been applied to solve this identication problem, they show di±culties when dealing with gaits at diferent speeds. For this reason, in this paper, a method to extract an average template from instances of the gait at diferent velocities is presented. This method has been tested with the gait signals of 34 subjects while walking at diferent motion speeds (slow, normal and fast) and it has shown to improve the performance of Euclidean distance and classical dynamic time warping.

A simplified spectral approachfor impedance-based damage identification of frp-strengthened rc beams

Hoy en día, el refuerzo y reparación de estructuras de hormigón armado mediante el pegado de bandas de polímeros reforzados con fibras (FRP) se emplea cada vez con más frecuencia a causa de sus numerosas ventajas. Sin embargo, las vigas reforzadas con esta técnica pueden experimentar un modo de fallo frágil a causa del despegue repentino de la banda de FRP a partir de una fisura intermedia. A pesar de su importancia, el número de trabajos que abordan el estudio de este mecanismo de fallo y su monitorización es muy limitado. Por ello, el desarrollo de metodologías capaces de monitorizar a largo plazo la adherencia de este refuerzo a las estructuras de hormigón e identificar cuándo se inicia el despegue de la banda constituyen un importante desafío a abordar. El principal objetivo de esta tesis es la implementación de una metodología fiable y efectiva, capaz de detectar el despegue de una banda de FRP en una viga de hormigón armado a partir de una fisura intermedia. Para alcanzar este objetivo se ha implementado un procedimiento de calibración numérica a partir de ensayos experimentales. Para ello, en primer lugar, se ha desarrollado un modelo numérico unidimensional simple y no costoso representativo del comportamiento de este tipo vigas de hormigón reforzadas con FRP, basado en un modelo de fisura discreta para el hormigón y el método de elementos espectrales. La formación progresiva de fisuras a flexion y el consiguiente despegue en la interface entre el hormigón y el FRP se formulan mediante la introducción de un nuevo elemento capaz de representar ambos fenómenos simultáneamente sin afectar al procedimiento numérico. Además, con el modelo propuesto, se puede obtener de una forma sencilla la respuesta dinámica en altas frecuencias de este tipo de estructuras, lo cual puede hacer muy útil su uso como herramienta de diagnosis y detección del despegue en su fase inicial mediante una monitorización de la variación de las características dinámicas locales de la estructura. Un método de evaluación no destructivo muy prometedor para la monitorización local de las estructuras es el método de la impedancia usando sensores-actuadores piezoeléctricos (PZT). La impedancia eléctrica de los sensores PZT se puede relacionar con la impedancia mecánica de las estructuras donde se encuentran adheridos Ya que la impedancia mecánica de una estructura se verá afectada por su deterioro, se pueden implementar indicadores de daño mediante una comparación del espectro de admitancia (inversa de la impedancia) a lo largo de distintas etapas durante el periodo de servicio de una estructura. Cualquier cambio en el espectro se podría interpretar como una variación en la integridad de la estructura. La impedancia eléctrica se mide a altas frecuencias con lo cual esta metodología debería ser muy sensible a la detección de estados de daño incipiente local, tal como se desea en la aplicación de este trabajo. Se ha implementado un elemento espectral PZT-FRP como extensión del modelo previamente desarrollado, con el objetivo de poder calcular numéricamente la impedancia eléctrica de sensores PZT adheridos a bandas de FRP sobre una viga de hormigón armado. El modelo, combinado con medidas experimentales captadas mediante sensores PZT, se implementa en el marco de una metodología de calibración de modelos para detectar cuantitativamente el despegue en la interfase entre una banda de FRP y una viga de hormigón. El procedimiento de optimización se resuelve empleando el método del enjambre cooperativo con un algoritmo bagging. Los resultados muestran una gran aproximación en la estimación del daño para el problema propuesto. Adicionalmente, se ha desarrollado también un método adaptativo para el mallado de elementos espectrales con el objetivo de localizar las zonas dañadas a partir de los resultados experimentales, el cual contribuye a aumentar la robustez y efectividad del método propuesto a la hora de identificar daños incipientes en su aparición inicial. Finalmente, se ha llevado a cabo un procedimiento de optimización multi-objetivo para detectar el despegue inicial en una viga de hormigón a escala real reforzada con FRP a partir de las impedancias captadas con una red de sensores PZT instrumentada a lo largo de la longitud de la viga. Cada sensor aporta los datos para definir cada una de las funciones objetivo que definen el procedimiento. Combinando el modelo previo de elementos espectrales con un algoritmo PSO multi-objetivo el procedimiento de detección de daño resultante proporciona resultados satisfactorios considerando la escala de la estructura y todas las incertidumbres características ligadas a este proceso. Los resultados obtenidos prueban la viabilidad y capacidad de los métodos antes mencionados y también su potencial en aplicaciones reales. Abstract Nowadays, the external bonding of fibre reinforced polymer (FRP) plates or sheets is increasingly used for the strengthening and retrofitting of reinforced concrete (RC) structures due to its numerous advantages. However, this kind of strengthening often leads to brittle failure modes being the most dominant failure mode the debonding induced by an intermediate crack (IC). In spite of its importance, the number of studies regarding the IC debonding mechanism and bond health monitoring is very limited. Methodologies able to monitor the long-term efficiency of bonding and successfully identify the initiation of FRP debonding constitute a challenge to be met. The main purpose of this thesisis the implementation of a reliable and effective methodology of damage identification able to detect intermediate crack debonding in FRP-strengthened RC beams. To achieve this goal, a model updating procedure based on numerical simulations and experimental tests has been implemented. For it, firstly, a simple and non-expensive one-dimensional model based on the discrete crack approach for concrete and the spectral element method has been developed. The progressive formation of flexural cracks and subsequent concrete-FRP interfacial debonding is formulated by the introduction of a new element able to represent both phenomena simultaneously without perturbing the numerical procedure. Furthermore, with the proposed model, high frequency dynamic response for these kinds of structures can also be obtained in a very simple and non-expensive way, which makes this procedure very useful as a tool for diagnoses and detection of debonding in its initial stage by monitoring the change in local dynamic characteristics. One very promising active non-destructive evaluation method for local monitoring is impedance-based structural health monitoring(SHM)using piezoelectric ceramic (PZT) sensor-actuators. The electrical impedance of the PZT can be directly related to the mechanical impedance of the host structural component where the PZT transducers are attached. Since the structural mechanical impedance will be affected by the presence of structural damage, comparisons of admittance (inverse of impedance) spectra at various times during the service period of the structure can be used as damage indicator. Any change in the spectra might be an indication of a change in the structural integrity. The electrical impedance is measured at high frequencies with which this methodology appears to be very sensitive to incipient damage in structural systems as desired for our application. Abonded-PZT-FRP spectral beam element approach based on an extension of the previous discrete crack approach is implemented in the calculation of the electrical impedance of the PZT transducer bonded to the FRP plates of a RC beam. This approach in conjunction with the experimental measurements of PZT actuator-sensors mounted on the structure is used to present an updating methodology to quantitatively detect interfacial debonding between a FRP strip and the host RC structure. The updating procedure is solved by using an ensemble particle swarm optimization approach with abagging algorithm, and the results demonstrate a big improvement for the performance and accuracy of the damage detection in the proposed problem. Additionally, an adaptive strategy of spectral element mesh has been also developed to detect damage location with experimental results, which shows the robustness and effectiveness of the proposed method to identify initial and incipient damages at its early stage. Lastly, multi-objective optimization has been carried out to detect debonding damage in a real scale FRP-strengthened RC beam by using impedance signatures. A net of PZT sensors is distributed along the beam to construct impedance-based multiple objectives under gradually induced damage scenario. By combining the spectral element model presented previously and an ensemble multi-objective PSO algorithm, the implemented damage detection process yields satisfactory predictions considering the scale and uncertainties of the structure. The obtained results prove the feasibility and capability of the aforementioned methods and also their potentials in real engineering applications.

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.