Speaker diarization: Segmentation and clustering of speeches

Speaker diarization is the process of sorting speeches according to the speaker. Diarization helps to search and retrieve what a certain speaker uttered in a meeting. Applications of diarization systemsextend to other domains than meetings, for example, lectures, telephone, television, and radio. Besides, diarization enhances the performance of several speech technologies such as speaker recognition, automatic transcription, and speaker tracking. Methodologies previously used in developing diarization systems are discussed. Prior results and techniques are studied and compared. Methods such as Hidden Markov Models and Gaussian Mixture Models that are used in speaker recognition and other speech technologies are also used in speaker diarization. The objective of this thesis is to develop a speaker diarization system in meeting domain. Experimental part of this work indicates that zero-crossing rate can be used effectively in breaking down the audio stream into segments, and adaptive Gaussian Models fit adequately short audio segments. Results show that 35 Gaussian Models and one second as average length of each segment are optimum values to build a diarization system for the tested data. Uniting the segments which are uttered by same speaker is done in a bottom-up clustering by a newapproach of categorizing the mixture weights.

Optimization and Measuring Techniques for Collect-and-Place Machines in Printed Circuit Board Industry

This thesis considers optimization problems arising in printed circuit board assembly. Especially, the case in which the electronic components of a single circuit board are placed using a single placement machine is studied. Although there is a large number of different placement machines, the use of collect-and-place -type gantry machines is discussed because of their flexibility and increasing popularity in the industry. Instead of solving the entire control optimization problem of a collect-andplace machine with a single application, the problem is divided into multiple subproblems because of its hard combinatorial nature. This dividing technique is called hierarchical decomposition. All the subproblems of the one PCB - one machine -context are described, classified and reviewed. The derived subproblems are then either solved with exact methods or new heuristic algorithms are developed and applied. The exact methods include, for example, a greedy algorithm and a solution based on dynamic programming. Some of the proposed heuristics contain constructive parts while others utilize local search or are based on frequency calculations. For the heuristics, it is made sure with comprehensive experimental tests that they are applicable and feasible. A number of quality functions will be proposed for evaluation and applied to the subproblems. In the experimental tests, artificially generated data from Markov-models and data from real-world PCB production are used. The thesis consists of an introduction and of five publications where the developed and used solution methods are described in their full detail. For all the problems stated in this thesis, the methods proposed are efficient enough to be used in the PCB assembly production in practice and are readily applicable in the PCB manufacturing industry.

Bayesian analysis of SEIR epidemic models

This thesis concerns the analysis of epidemic models. We adopt the Bayesian paradigm and develop suitable Markov Chain Monte Carlo (MCMC) algorithms. This is done by considering an Ebola outbreak in the Democratic Republic of Congo, former Zaïre, 1995 as a case of SEIR epidemic models. We model the Ebola epidemic deterministically using ODEs and stochastically through SDEs to take into account a possible bias in each compartment. Since the model has unknown parameters, we use different methods to estimate them such as least squares, maximum likelihood and MCMC. The motivation behind choosing MCMC over other existing methods in this thesis is that it has the ability to tackle complicated nonlinear problems with large number of parameters. First, in a deterministic Ebola model, we compute the likelihood function by sum of square of residuals method and estimate parameters using the LSQ and MCMC methods. We sample parameters and then use them to calculate the basic reproduction number and to study the disease-free equilibrium. From the sampled chain from the posterior, we test the convergence diagnostic and confirm the viability of the model. The results show that the Ebola model fits the observed onset data with high precision, and all the unknown model parameters are well identified. Second, we convert the ODE model into a SDE Ebola model. We compute the likelihood function using extended Kalman filter (EKF) and estimate parameters again. The motivation of using the SDE formulation here is to consider the impact of modelling errors. Moreover, the EKF approach allows us to formulate a filtered likelihood for the parameters of such a stochastic model. We use the MCMC procedure to attain the posterior distributions of the parameters of the SDE Ebola model drift and diffusion parts. In this thesis, we analyse two cases: (1) the model error covariance matrix of the dynamic noise is close to zero , i.e. only small stochasticity added into the model. The results are then similar to the ones got from deterministic Ebola model, even if methods of computing the likelihood function are different (2) the model error covariance matrix is different from zero, i.e. a considerable stochasticity is introduced into the Ebola model. This accounts for the situation where we would know that the model is not exact. As a results, we obtain parameter posteriors with larger variances. Consequently, the model predictions then show larger uncertainties, in accordance with the assumption of an incomplete model.

Estimation du modèle GARCH à changement de régimes et son utilité pour quantifier le risque de modèle dans les applications financières en actuariat

Le modèle GARCH à changement de régimes est le fondement de cette thèse. Ce modèle offre de riches dynamiques pour modéliser les données financières en combinant une structure GARCH avec des paramètres qui varient dans le temps. Cette flexibilité donne malheureusement lieu à un problème de path dependence, qui a empêché l'estimation du modèle par le maximum de vraisemblance depuis son introduction, il y a déjà près de 20 ans. La première moitié de cette thèse procure une solution à ce problème en développant deux méthodologies permettant de calculer l'estimateur du maximum de vraisemblance du modèle GARCH à changement de régimes. La première technique d'estimation proposée est basée sur l'algorithme Monte Carlo EM et sur l'échantillonnage préférentiel, tandis que la deuxième consiste en la généralisation des approximations du modèle introduites dans les deux dernières décennies, connues sous le nom de collapsing procedures. Cette généralisation permet d'établir un lien méthodologique entre ces approximations et le filtre particulaire. La découverte de cette relation est importante, car elle permet de justifier la validité de l'approche dite par collapsing pour estimer le modèle GARCH à changement de régimes. La deuxième moitié de cette thèse tire sa motivation de la crise financière de la fin des années 2000 pendant laquelle une mauvaise évaluation des risques au sein de plusieurs compagnies financières a entraîné de nombreux échecs institutionnels. À l'aide d'un large éventail de 78 modèles économétriques, dont plusieurs généralisations du modèle GARCH à changement de régimes, il est démontré que le risque de modèle joue un rôle très important dans l'évaluation et la gestion du risque d'investissement à long terme dans le cadre des fonds distincts. Bien que la littérature financière a dévoué beaucoup de recherche pour faire progresser les modèles économétriques dans le but d'améliorer la tarification et la couverture des produits financiers, les approches permettant de mesurer l'efficacité d'une stratégie de couverture dynamique ont peu évolué. Cette thèse offre une contribution méthodologique dans ce domaine en proposant un cadre statistique, basé sur la régression, permettant de mieux mesurer cette efficacité.

Improving sampling, optimization and feature extraction in Boltzmann machines

L’apprentissage supervisé de réseaux hiérarchiques à grande échelle connaît présentement un succès fulgurant. Malgré cette effervescence, l’apprentissage non-supervisé représente toujours, selon plusieurs chercheurs, un élément clé de l’Intelligence Artificielle, où les agents doivent apprendre à partir d’un nombre potentiellement limité de données. Cette thèse s’inscrit dans cette pensée et aborde divers sujets de recherche liés au problème d’estimation de densité par l’entremise des machines de Boltzmann (BM), modèles graphiques probabilistes au coeur de l’apprentissage profond. Nos contributions touchent les domaines de l’échantillonnage, l’estimation de fonctions de partition, l’optimisation ainsi que l’apprentissage de représentations invariantes. Cette thèse débute par l’exposition d’un nouvel algorithme d'échantillonnage adaptatif, qui ajuste (de fa ̧con automatique) la température des chaînes de Markov sous simulation, afin de maintenir une vitesse de convergence élevée tout au long de l’apprentissage. Lorsqu’utilisé dans le contexte de l’apprentissage par maximum de vraisemblance stochastique (SML), notre algorithme engendre une robustesse accrue face à la sélection du taux d’apprentissage, ainsi qu’une meilleure vitesse de convergence. Nos résultats sont présent ́es dans le domaine des BMs, mais la méthode est générale et applicable à l’apprentissage de tout modèle probabiliste exploitant l’échantillonnage par chaînes de Markov. Tandis que le gradient du maximum de vraisemblance peut-être approximé par échantillonnage, l’évaluation de la log-vraisemblance nécessite un estimé de la fonction de partition. Contrairement aux approches traditionnelles qui considèrent un modèle donné comme une boîte noire, nous proposons plutôt d’exploiter la dynamique de l’apprentissage en estimant les changements successifs de log-partition encourus à chaque mise à jour des paramètres. Le problème d’estimation est reformulé comme un problème d’inférence similaire au filtre de Kalman, mais sur un graphe bi-dimensionnel, où les dimensions correspondent aux axes du temps et au paramètre de température. Sur le thème de l’optimisation, nous présentons également un algorithme permettant d’appliquer, de manière efficace, le gradient naturel à des machines de Boltzmann comportant des milliers d’unités. Jusqu’à présent, son adoption était limitée par son haut coût computationel ainsi que sa demande en mémoire. Notre algorithme, Metric-Free Natural Gradient (MFNG), permet d’éviter le calcul explicite de la matrice d’information de Fisher (et son inverse) en exploitant un solveur linéaire combiné à un produit matrice-vecteur efficace. L’algorithme est prometteur: en terme du nombre d’évaluations de fonctions, MFNG converge plus rapidement que SML. Son implémentation demeure malheureusement inefficace en temps de calcul. Ces travaux explorent également les mécanismes sous-jacents à l’apprentissage de représentations invariantes. À cette fin, nous utilisons la famille de machines de Boltzmann restreintes “spike & slab” (ssRBM), que nous modifions afin de pouvoir modéliser des distributions binaires et parcimonieuses. Les variables latentes binaires de la ssRBM peuvent être rendues invariantes à un sous-espace vectoriel, en associant à chacune d’elles, un vecteur de variables latentes continues (dénommées “slabs”). Ceci se traduit par une invariance accrue au niveau de la représentation et un meilleur taux de classification lorsque peu de données étiquetées sont disponibles. Nous terminons cette thèse sur un sujet ambitieux: l’apprentissage de représentations pouvant séparer les facteurs de variations présents dans le signal d’entrée. Nous proposons une solution à base de ssRBM bilinéaire (avec deux groupes de facteurs latents) et formulons le problème comme l’un de “pooling” dans des sous-espaces vectoriels complémentaires.

Modeling High-Dimensional Audio Sequences with Recurrent Neural Networks

Cette thèse étudie des modèles de séquences de haute dimension basés sur des réseaux de neurones récurrents (RNN) et leur application à la musique et à la parole. Bien qu'en principe les RNN puissent représenter les dépendances à long terme et la dynamique temporelle complexe propres aux séquences d'intérêt comme la vidéo, l'audio et la langue naturelle, ceux-ci n'ont pas été utilisés à leur plein potentiel depuis leur introduction par Rumelhart et al. (1986a) en raison de la difficulté de les entraîner efficacement par descente de gradient. Récemment, l'application fructueuse de l'optimisation Hessian-free et d'autres techniques d'entraînement avancées ont entraîné la recrudescence de leur utilisation dans plusieurs systèmes de l'état de l'art. Le travail de cette thèse prend part à ce développement. L'idée centrale consiste à exploiter la flexibilité des RNN pour apprendre une description probabiliste de séquences de symboles, c'est-à-dire une information de haut niveau associée aux signaux observés, qui en retour pourra servir d'à priori pour améliorer la précision de la recherche d'information. Par exemple, en modélisant l'évolution de groupes de notes dans la musique polyphonique, d'accords dans une progression harmonique, de phonèmes dans un énoncé oral ou encore de sources individuelles dans un mélange audio, nous pouvons améliorer significativement les méthodes de transcription polyphonique, de reconnaissance d'accords, de reconnaissance de la parole et de séparation de sources audio respectivement. L'application pratique de nos modèles à ces tâches est détaillée dans les quatre derniers articles présentés dans cette thèse. Dans le premier article, nous remplaçons la couche de sortie d'un RNN par des machines de Boltzmann restreintes conditionnelles pour décrire des distributions de sortie multimodales beaucoup plus riches. Dans le deuxième article, nous évaluons et proposons des méthodes avancées pour entraîner les RNN. Dans les quatre derniers articles, nous examinons différentes façons de combiner nos modèles symboliques à des réseaux profonds et à la factorisation matricielle non-négative, notamment par des produits d'experts, des architectures entrée/sortie et des cadres génératifs généralisant les modèles de Markov cachés. Nous proposons et analysons également des méthodes d'inférence efficaces pour ces modèles, telles la recherche vorace chronologique, la recherche en faisceau à haute dimension, la recherche en faisceau élagué et la descente de gradient. Finalement, nous abordons les questions de l'étiquette biaisée, du maître imposant, du lissage temporel, de la régularisation et du pré-entraînement.

Information Exchange and Fusion in Dynamic and Heterogeneous Distributed Environments

Context awareness, dynamic reconfiguration at runtime and heterogeneity are key characteristics of future distributed systems, particularly in ubiquitous and mobile computing scenarios. The main contributions of this dissertation are theoretical as well as architectural concepts facilitating information exchange and fusion in heterogeneous and dynamic distributed environments. Our main focus is on bridging the heterogeneity issues and, at the same time, considering uncertain, imprecise and unreliable sensor information in information fusion and reasoning approaches. A domain ontology is used to establish a common vocabulary for the exchanged information. We thereby explicitly support different representations for the same kind of information and provide Inter-Representation Operations that convert between them. Special account is taken of the conversion of associated meta-data that express uncertainty and impreciseness. The Unscented Transformation, for example, is applied to propagate Gaussian normal distributions across highly non-linear Inter-Representation Operations. Uncertain sensor information is fused using the Dempster-Shafer Theory of Evidence as it allows explicit modelling of partial and complete ignorance. We also show how to incorporate the Dempster-Shafer Theory of Evidence into probabilistic reasoning schemes such as Hidden Markov Models in order to be able to consider the uncertainty of sensor information when deriving high-level information from low-level data. For all these concepts we provide architectural support as a guideline for developers of innovative information exchange and fusion infrastructures that are particularly targeted at heterogeneous dynamic environments. Two case studies serve as proof of concept. The first case study focuses on heterogeneous autonomous robots that have to spontaneously form a cooperative team in order to achieve a common goal. The second case study is concerned with an approach for user activity recognition which serves as baseline for a context-aware adaptive application. Both case studies demonstrate the viability and strengths of the proposed solution and emphasize that the Dempster-Shafer Theory of Evidence should be preferred to pure probability theory in applications involving non-linear Inter-Representation Operations.

Optimization-Based Robotic Manipulation for Safe Interaction with Human Operators

This thesis investigates a method for human-robot interaction (HRI) in order to uphold productivity of industrial robots like minimization of the shortest operation time, while ensuring human safety like collision avoidance. For solving such problems an online motion planning approach for robotic manipulators with HRI has been proposed. The approach is based on model predictive control (MPC) with embedded mixed integer programming. The planning strategies of the robotic manipulators mainly considered in the thesis are directly performed in the workspace for easy obstacle representation. The non-convex optimization problem is approximated by a mixed-integer program (MIP). It is further effectively reformulated such that the number of binary variables and the number of feasible integer solutions are drastically decreased. Safety-relevant regions, which are potentially occupied by the human operators, can be generated online by a proposed method based on hidden Markov models. In contrast to previous approaches, which derive predictions based on probability density functions in the form of single points, such as most likely or expected human positions, the proposed method computes safety-relevant subsets of the workspace as a region which is possibly occupied by the human at future instances of time. The method is further enhanced by combining reachability analysis to increase the prediction accuracy. These safety-relevant regions can subsequently serve as safety constraints when the motion is planned by optimization. This way one arrives at motion plans that are safe, i.e. plans that avoid collision with a probability not less than a predefined threshold. The developed methods have been successfully applied to a developed demonstrator, where an industrial robot works in the same space as a human operator. The task of the industrial robot is to drive its end-effector according to a nominal sequence of grippingmotion-releasing operations while no collision with a human arm occurs.

Probabilistic models to assist maintenance of multiple instruments

The paper discusses maintenance challenges of organisations with a huge number of devices and proposes the use of probabilistic models to assist monitoring and maintenance planning. The proposal assumes connectivity of instruments to report relevant features for monitoring. Also, the existence of enough historical registers with diagnosed breakdowns is required to make probabilistic models reliable and useful for predictive maintenance strategies based on them. Regular Markov models based on estimated failure and repair rates are proposed to calculate the availability of the instruments and Dynamic Bayesian Networks are proposed to model cause-effect relationships to trigger predictive maintenance services based on the influence between observed features and previously documented diagnostics

Characterizing PvARP, a novel Plasmodium vivax antigen

Background Plasmodium vivax continues to be the most widely distributed malarial parasite species in tropical and sub-tropical areas, causing high morbidity indices around the world. Better understanding of the proteins used by the parasite during the invasion of red blood cells is required to obtain an effective vaccine against this disease. This study describes characterizing the P. vivax asparagine-rich protein (PvARP) and examines its antigenicity in natural infection. Methods The target gene in the study was selected according to a previous in silico analysis using profile hidden Markov models which identified P. vivax proteins that play a possible role in invasion. Transcription of the arp gene in the P. vivax VCG-1 strain was here evaluated by RT-PCR. Specific human antibodies against PvARP were used to confirm protein expression by Western blot as well as its subcellular localization by immunofluorescence. Recognition of recombinant PvARP by sera from P. vivax-infected individuals was evaluated by ELISA. Results VCG-1 strain PvARP is a 281-residue-long molecule, which is encoded by a single exon and has an N-terminal secretion signal, as well as a tandem repeat region. This protein is expressed in mature schizonts and is located on the surface of merozoites, having an apparent accumulation towards their apical pole. Sera from P. vivax-infected patients recognized the recombinant, thereby suggesting that this protein is targeted by the immune response during infection.

GeneMCL in microarray analysis

Accurately and reliably identifying the actual number of clusters present with a dataset of gene expression profiles, when no additional information on cluster structure is available, is a problem addressed by few algorithms. GeneMCL transforms microarray analysis data into a graph consisting of nodes connected by edges, where the nodes represent genes, and the edges represent the similarity in expression of those genes, as given by a proximity measurement. This measurement is taken to be the Pearson correlation coefficient combined with a local non-linear rescaling step. The resulting graph is input to the Markov Cluster (MCL) algorithm, which is an elegant, deterministic, non-specific and scalable method, which models stochastic flow through the graph. The algorithm is inherently affected by any cluster structure present, and rapidly decomposes a graph into cohesive clusters. The potential of the GeneMCL algorithm is demonstrated with a 5730 gene subset (IGS) of the Van't Veer breast cancer database, for which the clusterings are shown to reflect underlying biological mechanisms. (c) 2005 Elsevier Ltd. All rights reserved.

Variable selection algorithm for the construction of MIMO operating point dependent neurofuzzy networks

An input variable selection procedure is introduced for the identification and construction of multi-input multi-output (MIMO) neurofuzzy operating point dependent models. The algorithm is an extension of a forward modified Gram-Schmidt orthogonal least squares procedure for a linear model structure which is modified to accommodate nonlinear system modeling by incorporating piecewise locally linear model fitting. The proposed input nodes selection procedure effectively tackles the problem of the curse of dimensionality associated with lattice-based modeling algorithms such as radial basis function neurofuzzy networks, enabling the resulting neurofuzzy operating point dependent model to be widely applied in control and estimation. Some numerical examples are given to demonstrate the effectiveness of the proposed construction algorithm.

Brain computer interface control via functional connectivity dynamics

The dynamics of inter-regional communication within the brain during cognitive processing – referred to as functional connectivity – are investigated as a control feature for a brain computer interface. EMDPL is used to map phase synchronization levels between all channel pair combinations in the EEG. This results in complex networks of channel connectivity at all time–frequency locations. The mean clustering coefficient is then used as a descriptive feature encapsulating information about inter-channel connectivity. Hidden Markov models are applied to characterize and classify dynamics of the resulting complex networks. Highly accurate levels of classification are achieved when this technique is applied to classify EEG recorded during real and imagined single finger taps. These results are compared to traditional features used in the classification of a finger tap BCI demonstrating that functional connectivity dynamics provide additional information and improved BCI control accuracies.

Bayesian analysis of skew-t multivariate null intercept measurement error model

The multivariate skew-t distribution (J Multivar Anal 79:93-113, 2001; J R Stat Soc, Ser B 65:367-389, 2003; Statistics 37:359-363, 2003) includes the Student t, skew-Cauchy and Cauchy distributions as special cases and the normal and skew-normal ones as limiting cases. In this paper, we explore the use of Markov Chain Monte Carlo (MCMC) methods to develop a Bayesian analysis of repeated measures, pretest/post-test data, under multivariate null intercept measurement error model (J Biopharm Stat 13(4):763-771, 2003) where the random errors and the unobserved value of the covariate (latent variable) follows a Student t and skew-t distribution, respectively. The results and methods are numerically illustrated with an example in the field of dentistry.

Semantic mapping using mobile robots

Robotic mapping is the process of automatically constructing an environment representation using mobile robots. We address the problem of semantic mapping, which consists of using mobile robots to create maps that represent not only metric occupancy but also other properties of the environment. Specifically, we develop techniques to build maps that represent activity and navigability of the environment. Our approach to semantic mapping is to combine machine learning techniques with standard mapping algorithms. Supervised learning methods are used to automatically associate properties of space to the desired classification patterns. We present two methods, the first based on hidden Markov models and the second on support vector machines. Both approaches have been tested and experimentally validated in two problem domains: terrain mapping and activity-based mapping.