Speech recognition systems have been implemented using a wide range of signal processing techniques including neuromorphic/biological inspired and Digital Signal Processing techniques. Neuromorphic/biologically inspired techniques, such as silicon cochlea models, are based on fairly simple yet highly parallel computation and/or computational units. While the area of digital signal processing (DSP) is based on block transforms and statistical or error minimization methods. Essential to each of these techniques is the first stage of extracting meaningful information from the speech signal, which is known as feature extraction. This can be done using biologically inspired techniques such as silicon cochlea models, or techniques beginning with a model of speech production and then trying to separate the the vocal tract response from an excitation signal. Even within each of these approaches, there are multiple techniques including cepstrum filtering, which sits under the class of Homomorphic signal processing, or techniques using FFT based predictive approaches. The underlying reality is there are multiple techniques that have attacked the problem in speech recognition but the problem is still far from being solved. The techniques that have shown to have the best recognition rates involve Cepstrum Coefficients for the feature extraction and Hidden-Markov Models to perform the pattern recognition. The presented research develops an analog system based on programmable analog array technology that can perform the initial stages of auditory feature extraction and recognition before passing information to a digital signal processor. The goal being a low power system that can be fully contained on one or more integrated circuit chips. Results show that it is possible to realize advanced filtering techniques such as Cepstrum Filtering and Vector Quantization in analog circuitry. Prior to this work, previous applications of analog signal processing have focused on vision, cochlea models, anti-aliasing filters and other single component uses. Furthermore, classic designs have looked heavily at utilizing op-amps as a basic core building block for these designs. This research also shows a novel design for a Hidden Markov Model (HMM) decoder utilizing circuits that take advantage of the inherent properties of subthreshold transistors and floating-gate technology to create low-power computational blocks.

This thesis analyzes the human action recognition problem. Human actions are modeled as a time evolving temporal texture. Gabor filters, which are proved to be a robust 2D texture representation tool by detecting spatial points with high variation, is extended to 3D domain to capture motion texture features. A well known filtering algorithm and a recent unsupervised clustering algorithm, the Genetic Chromodynamics, are combined to select salient spatio-temporal features of the temporal texture and to segment the activity sequence into temporal texture primitives. Each activity sequence is represented as a composition of temporal texture primitives with its salient spatio-temporal features, which are also the symbols of our codebook. To overcome temporal variation between different performances of the same action, a Profile Hidden Markov Model is applied with Viterbi Path Counting (ensemble training). Not only parameters and structure but also codebook is learned during training.

In the Internet era, more and more courses are taught through a course management system (CMS) or learning management system (LMS). In an asynchronous virtual learning environment, an instructor has the need to beware the progress of discussions in forums, and may intervene if ecessary in order to facilitate students¡¦ learning. This research proposes a discussion forum activity flow tracking system, called FAFT (Forum Activity Flow Tracer), to utomatically monitor the discussion activity flow of threaded forum postings in CMS/LMS. As CMS/LMS is getting popular in facilitating learning activities, the proposedFAFT can be used to facilitate instructors to identify students¡¦ interaction types in discussion forums. FAFT adopts modern data/text mining techniques to discover the patterns of forum discussion activity flows, which can be used for instructors to facilitate the online learning activities. FAFT consists of two subsystems: activity classification (AC) and activity flow discovery (AFD). A posting can be perceived as a type of announcement, questioning, clarification, interpretation, conflict, or assertion. AC adopts a cascade model to classify various activitytypes of posts in a discussion thread. The empirical evaluation of the classified types from a repository of postings in earth science on-line courses in a senior high school shows that AC can effectively facilitate the coding rocess, and the cascade model can deal with the imbalanced distribution nature of discussion postings. AFD adopts a hidden Markov model (HMM) to discover the activity flows. A discussion activity flow can be presented as a hidden Markov model (HMM) diagram that an instructor can adopt to predict which iscussion activity flow type of a discussion thread may be followed. The empirical results of the HMM from an online forum in earth science subject in a senior high school show that FAFT can effectively predict the type of a discussion activity flow. Thus, the proposed FAFT can be embedded in a course management system to automatically predict the activity flow type of a discussion thread, and in turn reduce the teachers¡¦ loads on managing online discussion forums.

In this thesis, I have studied CMOS dendrites, implemented them on a reconfigurable analog platform and modeled them using MATLAB Simulink. The dendrite model was further used to build a computational model. I implemented a Hidden Markov Model (HMM) classifier to build a simple YES/NO wordspotter. I also discussed the inter-relation between neural systems, CMOS transistors and HMM networks. The physical principles behind the operation of silicon devices and biological structures are similar. Hence silicon devices can be used to emulate biological structures like dendrites. Dendrites are a branched, conductive medium which connect a neurons synapses to its soma. Dendrites were previously believed to be like wires in neural networks. However, recent research suggests that they have computational power. We can emulate dendrites using transistors in the Field Programmable Analog Array (FPAA). Our lab has built the Reconfigurable Analog Signal Processor (RASP) family of FPAAs which was used for the experiments. I analytically compared the mathematical model of dendrites to our model in silicon. The mathematical model based on the device physics of the silicon devices was then used to simulate dendrites in Simulink. An automated tool, sim2spice was then used to convert the Simulink model into a SPICE netlist, such that it can be implemented on a FPAA. This is an easier tool to use for DSP and Neuromorphic engineers who's primary areas of expertise isn't circuit design.

Είναι ενδιαφέρουσα αλλά συνάμα δύσκολη η ανάλυση ιατρικών εικόνων, επειδή υπάρχουν πολύ μικρές διακυμάνσεις και μεγάλος όγκος δεδομένων για επεξεργασία. Είναι αρκετά δύσκολο να αναπτυχθεί ένα αυτοματοποιημένο σύστημα αναγνώρισης, το οποίο θα μπορούσε να επεξεργάζεται μεγάλο όγκο πληροφοριών των ασθενών και να παρέχει μια σωστή εκτίμηση. Στην ιατρική, η συμβατική διαγνωστική μέθοδος για εικόνες MR γονάτου για αναγνώριση ανωμαλιών, είναι από την επίβλεψη έμπειρων ιατρών. Η τεχνική της ασαφούς λογικής είναι πιο ακριβής, αλλά αυτό εξαρτάται πλήρως από τη γνώση των εμπειρογνωμόνων, η οποία μπορεί να μην είναι πάντα διαθέσιμη. Στη παρούσα εργασία, τμηματοποιούμε την MR εικόνα του γονάτου με την τεχνική Mean Shift, αναγνωρίζουμε τα κύρια μέρη με τη βοήθεια των ΗΜRF και τέλος εκπαιδεύουμε ταξινομητή ANFIS. Η απόδοση του ταξινομητή ANFIS αξιολογήθηκε όσον αφορά την απόδοση της εκαπαίδευσης και της ακρίβειας ταξινόμησης. Επιβεβαιώθηκε ότι ο ταξινομητής είχε μεγάλη ακρίβεια στην ανίχνευση ανωμαλιών στις ακτινογραφίες Στην εργασία αυτή περιγράφεται η προτεινόμενη στρατηγική για την διάγνωση ανωμαλιών στις εικόνες μαγνητικής τομογραφίας γόνατος. / It is a challenging task to analyze medical images because there are very minute variations & larger data set for analysis. It is a quite difficult to develop an automated recognition system which could process on a large information of patient and provide a correct estimation. The conventional method in medicine for knee MR images classification and diseases detection is by human inspection. Fuzzy logic technique is more accurate but it fully depends on expert knowledge, which may not always available. Here we extract the feature using Mean Shift segmentation and region recognition with HMRF and after that training using the ANFIS tool. The performance of the ANFIS classifier was evaluated in terms of training performance and classification accuracy. Here the result confirmed that the proposed ANFIS classifier with high accuracy in detecting the knee diseases. This work describes the proposed strategy to medical image classification of patient’s MRI scan images of the knee.

Tracking and understanding human gait is an important step towards improving elderly mobility and safety. This thesis presents a vision-based tracking system that estimates the 3D pose of a wheeled walker user's lower limbs with cameras mounted on the moving walker. The tracker estimates 3D poses from images of the lower limbs in the coronal plane in a dynamic, uncontrolled environment. It employs a probabilistic approach based on particle filtering with three different camera setups: a monocular RGB camera, binocular RGB cameras, and a depth camera. For the RGB cameras, observation likelihoods are designed to compare the colors and gradients of each frame with initial templates that are manually extracted. Two strategies are also investigated for handling appearance change of tracking target: increasing number of templates and using different representations of colors. For the depth camera, two observation likelihoods are developed: the first one works directly in the 3D space, while the second one works in the projected image space. Experiments are conducted to evaluate the performance of the tracking system with different users for all three camera setups. It is demonstrated that the trackers with the RGB cameras produce results with higher error as compared to the depth camera, and the strategies for handling appearance change improve tracking accuracy in general. On the other hand, the tracker with the depth sensor successfully tracks the 3D poses of users over the entire video sequence and is robust against unfavorable conditions such as partial occlusion, missing observations, and deformable tracking target.

One of the biggest challenges in speech recognition today is its use on a daily basis, in which distortion and noise in the environment are present and hinder this task. In the last thirty years, hundreds of methods for noise-robust recognition were proposed, each with its own advantages and disadvantages. In this thesis, the use of Convolutional Neural Networks (CNN) as acoustic models in automatic speech recognition systems (ASR) is proposed as an alternative to the classical recognition methods based on Hidden Markov Models (HMM) without any noise-robust method applied. Experiments were performed with a audio set modified by additive and natural noises, and showed that the presented method reduces the Equal Error Rate (EER) and improves the acuracy of speech recognition in noisy environments when compared to traditional models of classifiation, indicating the robustness of the approach. / Um dos maiores desafios no reconhecimento de fala atualmente é usá-lo no contexto diário, no qual distorções no sinal da fala e ruídos no ambiente estão presentes e re- duzem a qualidade do reconhecimento. Nos últimos trinta anos, centenas de métodos para reconhecimento robusto ao ruído foram propostos, cada um com suas vantagens e desvantagens. Este trabalho propõe o uso de uma rede neural convolucional no papel de modelo acústico em sistemas de reconhecimento automático de fala,como uma alter- nativa ao métodos clássicos de reconhecimento baseado em modelos ocultos de Markov (HMM, do inglês, Hidden Markov Models) sem a aplicação de um método robusto ao ruído. Experimentos foram realizados com áudios modi ficados com ruídos aditivos e reais, e mostraram que o método proposto reduz o Equal Error Rate (EER) e aumenta a acurácia da classificação de comando de voz quando comparado a modelos tradicionais de classificação, evidenciando a robustez da abordagem apresentada.

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / This work is a study of the relationship between the intrinsic dimension of feature vectors applied to the classification of video signals in order to perform lip reading. In pattern recognition tasks, the extraction of relevant features is crucial for a good performance of the classifiers. The starting point of this work was the reproduction of the work of J.R. Movellan [1], which classifies lips gestures with HMM using only the video signal from the Tulips1 database. The database consists of videos of volunteers’ mouths while they utter the first 4 numerals in English. The original work uses feature vectors of high dimensionality in relation to the size of the database. Consequently, the adjustment of HMM classifiers has become problematic and the maximum accuracy was only 66.67%. Alternative strategies for feature extraction and classification schemes were proposed in order to analyze the influence of the intrinsic dimension in the performance of classifiers. The best solution, in terms of results, achieved an accuracy of approximately 83%. / Este trabalho é um estudo da relação entre a dimensão intrínseca de vetores de características aplicados à classificação de sinais de vídeo no intuito de realizar-se a leitura labial. Nas tarefas de reconhecimento de padrões, a extração de características relevantes é crucial para um bom desempenho dos classificadores. O ponto de partida deste trabalho foi a reprodução do trabalho de J.R. Movellan [1], que realiza a classificação de gestos labiais com HMM na base de dados Tulips1, utilizando somente o sinal de vídeo. A base é composta por vídeos das bocas de voluntários enquanto esses pronunciam os primeiros 4 numerais em inglês. O trabalho original utiliza vetores de características de dimensão muito alta em relação ao tamanho da base. Consequentemente, o ajuste de classificadores HMM se tornou problemático e só se alcançou 66,67% de acurácia. Estratégias de extração de características e esquemas de classificação alternativos foram propostos, a fim de analisar a influência da dimensão intrínseca no desempenho de classificadores. A melhor solução, em termos de resultados, obteve uma acurácia de aproximadamente 83%. / São Cristóvão, SE

Orientador: Fabio Violaro / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-11T22:49:44Z (GMT). No. of bitstreams: 1 Selmini_AntonioMarcos_D.pdf: 2404244 bytes, checksum: d7fcd0828f3157c595a0e3426b4a7eb0 (MD5) Previous issue date: 2008 / Resumo: A demanda por uma segmentação automática de fala confiável vem crescendo e exigindo pesquisas para suportar o desenvolvimento de sistemas que usam fala para uma interação homem-máquina. Neste contexto, este trabalho relata o desenvolvimento e avaliação de um sistema para segmentação automática de fala usando o algoritmo de Viterbi e refinamento das fronteiras de segmentação baseado nas características fonético-acústicas das classes fonéticas. As subunidades fonéticas (dependentes de contexto) são representadas com Modelos Ocultos de Markov (HMM - Hidden Markov Models). Cada fronteira estimada pelo algoritmo de Viterbi é refinada usando características acústicas dependentes de classes de fones, uma vez que a identidade dos fones do lado direito e esquerdo da fronteira considerada é conhecida. O sistema proposto foi avaliado usando duas bases dependentes de locutor do Português do Brasil (uma masculina e outra feminina) e também uma base independente de locutor (TIMIT). A avaliação foi realizada comparando a segmentação automática com a segmentação manual. Depois do processo de refinamento, um ganho de 29% nas fronteiras com erro de segmentação abaixo de 20 ms foi obtido para a base de fala dependente de locutor masculino do Português Brasileiro. / Abstract: The demand for reliable automatic speech segmentation is increasing and requiring additional research to support the development of systems that use speech for man-machine interface. In this context, this work reports the development and evaluation of a system for automatic speech segmentation using Viterbi's algorithm and a refinement of segmentation boundaries based on acoustic-phonetic features. Phonetic sub-units (context-dependent phones) are modeled with HMM (Hidden Markov Models). Each boundary estimated by Viterbi's algorithm is refined using class-dependent acoustic features, as the identity of the phones on the left and right side of the considered boundary is known. The proposed system was evaluated using two speaker dependent Brazilian Portuguese speech databases (one male and one female speaker), and a speaker independent English database (TIMIT). The evaluation was carried out comparing automatic against manual segmentation. After the refinement process, an improvement of 29% in the percentage of segmentation errors below 20 ms was achieved for the male speaker dependent Brazilian Portuguese speech database. / Doutorado / Telecomunicações e Telemática / Doutor em Engenharia Elétrica

The vital signs monitoring data of an infant receiving intensive care are a rich source of information about its health condition. One major concern about the state of health of such patients is the onset of neonatal sepsis, a life-threatening bloodstream infection. As early signs are subtle and current diagnosis procedures involve slow laboratory testing, sepsis detection based on the monitored physiological dynamics is a clinically significant task. This challenging problem can be thoroughly modelled as real-time inference within a machine learning framework. In this thesis, we develop probabilistic dynamical models centred around the goal of providing useful predictions about the onset of neonatal sepsis. This research is characterised by the careful incorporation of domain knowledge for the purpose of extracting the infant’s true physiology from the monitoring data. We make two main contributions. The first one is the formulation of sepsis detection as learning and inference in an Auto-Regressive Hidden Markov Model (AR-HMM). The model investigates the extent to which physiological events observed in the patient’s monitoring traces could be used for the early detection of neonatal sepsis. In addition, the proposed approach involves exact marginalisation over missing data at inference time. When applying the ARHMM on a real-world dataset, we found that it can produce effective predictions about the onset of sepsis. Second, both sepsis and clinical event detection are formulated as learning and inference in a Hierarchical Switching Linear Dynamical System (HSLDS). The HSLDS models dynamical systems where complex interactions between modes of operation can be represented as a twolevel hidden discrete hierarchical structure. For neonatal condition monitoring, the lower layer models clinical events and is controlled by upper layer variables with semantics sepsis/nonsepsis. The model parameterisation and estimation procedures are adapted to the specifics of physiological monitoring data. We demonstrate that the performance of the HSLDS for the detection of sepsis is not statistically different from the AR-HMM, despite the fact that the latter model is given “ground truth” annotations of the patient’s physiology.