Ever since the birth of the telephony system, the problem with echoes, arising from impedance mismatch in 2/4-wire hybrids, or acoustic echoes where a loudspeaker signal is picked up by a closely located microphone, has been ever present. The removal of these echoes is crucial in order to achieve an acceptable audio quality for conversation. Today, the perhaps most common way for echo removal is through cancellation, where an adaptive filter is used to produce an estimated replica of the echo which is then subtracted from the echo-infested signal. Echo cancellation in practice requires extensive control of the filter adaptation process in order to obtain as rapid convergence as possible while also achieving robustness towards disturbances. Moreover, despite the rapid advancement in the computational capabilities of modern digital signal processors there is a constant demand for low-complexity solutions that can be implemented using low power and low cost hardware. This thesis presents low-complexity solutions for echo cancellation related to both the actual filter adaptation process itself as well as for controlling the adaptation process in order to obtain a robust system. Extensive simulations and evaluations using real world recorded signals are used to demonstrate the performance of the proposed solutions.

Cette thèse traite - en trois essais - de problèmes de choix de portefeuille en situation d’information partielle, thématique que nous présentons dans une courte introduction. Les essais développés abordent chacun une particularité de cette problématique. Le premier (coécrit avec M. Jeanblanc et V. Lacoste) traite la question du choix de la stratégie optimale pour un problème de maximisation d’utilité terminale lorsque l’évolution des prix est modélisée par un processus de Itô-Lévy dont la tendance et l’intensité des sauts ne sont pas observées. L’approche consiste à réécrire le problème initial comme un problème réduit dans la filtration engendrée par les prix. Cela nécessite la dérivation des équations de filtrage non-linéaire, que nous développons pour un processus de Lévy. Le problème est ensuite résolu en utilisant la programmation dynamique par les équations de Bellman et de HJB. Le second essai aborde dans un cadre gaussien la question du coût de l’incertitude, que nous définissons comme la différence entre les stratégies optimales (ou les richesses maximales) d’un agent parfaitement informé et d’un agent partiellement informé. Les propriétés de ce coût de l’information sont étudiées dans le cadre des trois formes standard de fonctions d’utilités et des exemples numériques sont présentés. Enfin, le troisième essai traite la question du choix de portefeuille quand l’information sur les prix de marché n’est disponible qu’à des dates discrètes et aléatoires. Cela revient à supposer que la dynamique des prix suit un processus marqué. Dans ce cadre, nous développons les équations de filtrage et réécrivons le problème initial dans sa forme réduite dans la filtration discrète des prix. Les stratégies optimales sont ensuite calculées en utilisant le calcul de Malliavin pour des mesures aléatoires et une extension de la formule de Clark-Ocone-Haussman est à cette fin présentée. / This thesis deals - in three essays - with problems of choice of portfolio in situation of partial information, thematic that we present in a short introduction. The tests developed each address a particularity of this problem. The first (co-written with M. Jeanblanc and V. Lacoste) deals with the choice of the optimal strategy for a terminal utility maximization problem when the evolution of prices is modeled by an Itô- Lévy process whose trend and the intensity of the jumps are not observed. The approach is to rewrite the initial problem as a reduced problem in price-driven fi ltration. This requires the derivation of nonlinear filtering equations, which we develop for a Lévy process. The problem is then solved using dynamic programming by the Bellman and HJB equations. The second essay tackles the question of the cost of uncertainty in a Gaussian framework, which we de fi ne as the di ff erence between the optimal strategies (or the maximum wealth) of a fully informed agent and a partially informed agent. The properties of this information cost are studied in the context of the three standard forms of utility functions and numericalexamples are presented. Finally, the third essay addresses the issue of portfolio choice when market price information is only available on discrete and random dates. This amounts to assuming that price dynamics follow a marked process. In this framework, we develop fi ltering equations and rewrite the initialproblem in its reduced form in discrete price fi ltration. The optimal strategies are then calculated using Malliavin's computation for random measurements and an extension of the Clark-Ocone-Haussman formula is for this purpose presented.

[pt] Este trabalho tem como finalidade explorar uma técnica de redução do erro de estimativas da resposta de freqüência discreta do canal geradas por símbolos piloto em sistemas de transmissão OFDM (Orthogonal Frequency Division Multiplexing). Nesta técnica, uma transformação linear projeta o vetor que contem as estimativas obtidas inicialmente no subespaço em que a verdadeira resposta de freqüência do canal tem que estar, resultando em uma redução da variância do erro das estimativas. A aplicação conjunta desta técnica com filtragem adaptativa para a estimação da resposta de freqüência do canal também está no contexto desta dissertação. Os resultados dos experimentos são analisados em termos da taxa de erro de bit média obtida e da convergência dos algoritmos adaptaivos empregados nas etapas de estimação de canal no receptor. / [en] This work a technique for error reduction in estimates of the discrete channel frequency response obtained with aid of pilot symbols in OFDM (Orthogonal Frequency Division Multiplexing) transmission systems. In this technique projects the vector that contains the initial discrete channel frequency response estimate is projected into the subspace where the true channel frequency response has to lye, yielding a new channel estimate with a reduced error variance. The joint application of this technique with adaptive filtering for channel estimation is also developed herein. The performance of the proposed methods is analyzed in terms of the mean bit error rate achieved and of the convergence of the adaptive channel estimation algorithms used in the receiver.

Luo Yilun. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2006. / Includes bibliographical references (leaves 87-91). / Abstracts in English and Chinese. / Chapter 1. --- Introduction --- p.1 / Chapter 1.1. --- Organization --- p.3 / Chapter 2. --- Architecture of MAG-μIMU --- p.5 / Chapter 2.1. --- Hardware for Attitude Filter --- p.5 / Chapter 2.2. --- Handwriting Recording for a Digital Writing Instrument --- p.7 / Chapter 3. --- Inertial Tracking for Handwriting --- p.9 / Chapter 3.1. --- Spatial Descriptions and Transformations --- p.9 / Chapter 3.1.1. --- Vector Description and Position of a Frame --- p.9 / Chapter 3.1.2. --- Coordinate Transformation and Orientation of a Frame --- p.10 / Chapter 3.1.3. --- Kinematics for Digital Writing Instruments --- p.12 / Chapter 3.1.4. --- Vector Rotation --- p.16 / Chapter 3.2. --- Euler Angles for Rotation in Space --- p.17 / Chapter 3.3. --- Euler Angles Attitude Kinematics --- p.19 / Chapter 3.4. --- Singular Problem --- p.19 / Chapter 4. --- Attitude in Quaternion --- p.22 / Chapter 4.1. --- Quaternion Operations --- p.22 / Chapter 4.1.1. --- Quaternion Conjugate --- p.23 / Chapter 4.1.2. --- Quaternion Norm --- p.24 / Chapter 4.1.3. --- Quaternion Inverse --- p.24 / Chapter 4.2. --- Orientation Description in Quaternion --- p.24 / Chapter 4.3. --- Attitude Kinematics in Quaternion --- p.25 / Chapter 5. --- Kalman Filter --- p.27 / Chapter 5.1. --- Time Update --- p.28 / Chapter 5.2. --- Measurement Update --- p.29 / Chapter 5.2.1. --- Maximum a Posterior Probability --- p.29 / Chapter 5.2.2. --- Batch Least-Square Estimation --- p.31 / Chapter 5.2.3. --- Measurement Update in Kalman Filter --- p.34 / Chapter 5.3. --- Kalman Filter Summary --- p.36 / Chapter 6. --- Extended Kalman Filter --- p.38 / Chapter 7. --- Attitude Extended Kalman Filter --- p.41 / Chapter 7.1. --- Time Update Model --- p.41 / Chapter 7.1.1. --- Attitude Strapdown Theory for a Quaternion --- p.41 / Chapter 7.1.2. --- Error Model for Time Update --- p.42 / Chapter 7.2. --- Measurement Update Model --- p.43 / Chapter 7.2.1. --- Error Model for the Measurement Update --- p.45 / Chapter 7.3. --- Summary --- p.46 / Chapter 8. --- Experiment Results --- p.47 / Chapter 8.1. --- Experiment for Attitude EKF based on MAG-μIMU --- p.47 / Chapter 8.1.1. --- Simulation Test --- p.48 / Chapter 8.1.2. --- Experiment Test --- p.49 / Chapter 8.2. --- Writing Application based on Attitude EKF Compensation --- p.52 / Chapter 8.2.1. --- Stroke Segment Kalman Filter --- p.54 / Chapter 8.2.2. --- Zero Velocity Compensation --- p.58 / Chapter 8.2.3. --- Complementary Attitude EKF for Writing Experiment --- p.60 / Chapter 9. --- Future Work --- p.73 / Chapter 9.1. --- Unscented Kalman Filter --- p.73 / Chapter 9.1.1. --- Least-square Estimator Structure --- p.73 / Chapter 9.1.2. --- Unscented Transform --- p.74 / Chapter 9.1.3. --- Unscented Kalman Filter --- p.76 / Chapter 9.2. --- Experiment Result --- p.81 / Chapter 10. --- Conclusion --- p.85 / Chapter 10.1. --- Attitude Extended Kalman Filter --- p.85 / Chapter 10.2. --- Complementary Attitude EKF --- p.85 / Chapter 10.3. --- Unscented Kalman Filter --- p.86 / Chapter 10.4. --- Future Work --- p.86 / Bibliography --- p.87 / Appendix A --- p.92

The goal of the thesis is to extend the kernel methods to matrix factorization(MF) for collaborative ltering(CF). In current literature, MF methods usually assume that the correlated data is distributed on a linear hyperplane, which is not always the case. The best known member of kernel methods is support vector machine (SVM) on linearly non-separable data. In this thesis, we apply kernel methods on MF, embedding the data into a possibly higher dimensional space and conduct factorization in that space. To improve kernelized matrix factorization, we apply multi-kernel learning methods to select optimal kernel functions from the candidates and introduce L2-norm regularization on the weight learning process. In our empirical study, we conduct experiments on three real-world datasets. The results suggest that the proposed method can improve the accuracy of the prediction surpassing state-of-art CF methods.

Thesis (M.S.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 1981. / MICROFICHE COPY AVAILABLE IN ARCHIVES AND ENGINEERING. / Includes bibliographical references. / by Lawrence Charles Vallot. / M.S.

Thesis. 1976. M.S.--Massachusetts Institute of Technology. Dept. of Electrical Engineering and Computer Science. / Microfiche copy available in Archives and Engineering. / Bibliography: leaves 145-148. / by Stuart C. Wright. / M.S.

Wang, Dingyan. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2011. / Includes bibliographical references (p. 97-110). / Abstracts in English and Chinese. / Abstract --- p.ii / Acknowledgement --- p.vi / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Overview --- p.1 / Chapter 1.2 --- Motivations --- p.6 / Chapter 1.3 --- Contributions --- p.9 / Chapter 1.4 --- Organizations of This Thesis --- p.11 / Chapter 2 --- Background --- p.14 / Chapter 2.1 --- Label Propagation Learning Framework --- p.14 / Chapter 2.1.1 --- Graph-based Semi-supervised Learning --- p.14 / Chapter 2.1.2 --- Green's Function Learning Framework --- p.16 / Chapter 2.2 --- Recommendation Methods --- p.19 / Chapter 2.2.1 --- Traditional Memory-based Methods --- p.19 / Chapter 2.2.2 --- Traditional Model-based Methods --- p.20 / Chapter 2.2.3 --- Label Propagation Recommendation Models --- p.22 / Chapter 2.2.4 --- Latent Feature Recommendation Models . --- p.24 / Chapter 2.2.5 --- Social Recommendation Models --- p.25 / Chapter 2.2.6 --- Tag-based Recommendation Models --- p.25 / Chapter 3 --- Recommendation with Latent Features --- p.28 / Chapter 3.1 --- Motivation and Contributions --- p.28 / Chapter 3.2 --- Item Graph --- p.30 / Chapter 3.2.1 --- Item Graph Definition --- p.30 / Chapter 3.2.2 --- Item Graph Construction --- p.31 / Chapter 3.3 --- Label Propagation Recommendation Model with Latent Features --- p.33 / Chapter 3.3.1 --- Latent Feature Analysis --- p.33 / Chapter 3.3.2 --- Probabilistic Matrix Factorization --- p.35 / Chapter 3.3.3 --- Similarity Consistency Between Global and Local Views (SCGL) --- p.39 / Chapter 3.3.4 --- Item-based Green's Function Recommendation Based on SCGL --- p.41 / Chapter 3.4 --- Experiments --- p.41 / Chapter 3.4.1 --- Dataset --- p.43 / Chapter 3.4.2 --- Baseline Methods --- p.43 / Chapter 3.4.3 --- Metrics --- p.45 / Chapter 3.4.4 --- Experimental Procedure --- p.45 / Chapter 3.4.5 --- Impact of Weight Parameter u --- p.46 / Chapter 3.4.6 --- Performance Comparison --- p.48 / Chapter 3.5 --- Summary --- p.50 / Chapter 4 --- Recommendation with Social Network --- p.51 / Chapter 4.1 --- Limitation and Contributions --- p.51 / Chapter 4.2 --- A Social Recommendation Framework --- p.55 / Chapter 4.2.1 --- Social Network --- p.55 / Chapter 4.2.2 --- User Graph --- p.57 / Chapter 4.2.3 --- Social-User Graph --- p.59 / Chapter 4.3 --- Experimental Analysis --- p.60 / Chapter 4.3.1 --- Dataset --- p.61 / Chapter 4.3.2 --- Metrics --- p.63 / Chapter 4.3.3 --- Experiment Setting --- p.64 / Chapter 4.3.4 --- Impact of Control Parameter u --- p.65 / Chapter 4.3.5 --- Performance Comparison --- p.67 / Chapter 4.4 --- Summary --- p.69 / Chapter 5 --- Recommendation with Tags --- p.71 / Chapter 5.1 --- Limitation and Contributions --- p.71 / Chapter 5.2 --- Tag-Based User Modeling --- p.75 / Chapter 5.2.1 --- Tag Preference --- p.75 / Chapter 5.2.2 --- Tag Relevance --- p.78 / Chapter 5.2.3 --- User Interest Similarity --- p.80 / Chapter 5.3 --- Tag-Based Label Propagation Recommendation --- p.83 / Chapter 5.4 --- Experimental Analysis --- p.84 / Chapter 5.4.1 --- Douban Dataset --- p.85 / Chapter 5.4.2 --- Experiment Setting --- p.86 / Chapter 5.4.3 --- Metrics --- p.87 / Chapter 5.4.4 --- Impact of Tag and Rating --- p.88 / Chapter 5.4.5 --- Performance Comparison --- p.90 / Chapter 5.5 --- Summary --- p.92 / Chapter 6 --- Conclusions and Future Work --- p.94 / Chapter 6.0.1 --- Conclusions --- p.94 / Chapter 6.0.2 --- Future Work --- p.96 / Bibliography --- p.97

This thesis examines the legal and governance issues presented by internet blocking (“filtering”) systems through the use of the United Kingdom’s Cleanfeed system as a national case study. The Cleanfeed system – which aims to block access to child abuse images – has been influential both domestically and internationally but has been the subject of relatively little sustained scrutiny in the literature. Using a mixed doctrinal and empirical methodology this work discusses the evolution of Cleanfeed and considers the way in which government pressure has led to a private body without any express legislative basis (the Internet Watch Foundation) being given the power to control what UK internet users can view. The thesis argues that the Cleanfeed system sits at the intersection of three distinct trends – the use of architectural regulation, regulation through intermediaries and self-regulation – which individually and collectively present significant risks for freedom of expression and good governance online. It goes on to identify and examine the fundamental rights norms and governance standards which should apply to internet blocking and tests the system against them, arguing in particular that Cleanfeed fails to meet the requirements developed by the European Court of Human Rights under Articles 6 and 10 ECHR. It considers the extent to which Cleanfeed might be made amenable to these principles through the use of judicial review or actions under the Human Rights Act 1998 and concludes that the diffuse structure of the system and the limited availability of horizontal effect against private bodies will leave significant aspects beyond the effective reach of the courts. This work also assesses claims that the Cleanfeed system is a proof of concept which should be extended so as to block other material considered objectionable (such as websites which “glorify terrorism”). It argues that the peculiar features of the system mean that it represents a best case scenario and does not support blocking of other types of content which are significantly more problematic. The thesis concludes by considering proposals for reform of the Cleanfeed system and the extent to which greater public law oversight might undermine the desirable features associated with self-regulation.

This dissertation studies the factors that negatively impact the accuracy of the collaborative filtering recommendation systems based on nonnegative matrix factorization (NMF). The keystone in the recommendation system is the rating that expresses the user's opinion about an item. One of the most significant issues in the recommendation systems is the lack of ratings. This issue is called "cold-start" issue, which appears clearly with New-Users who did not rate any item and New-Items, which did not receive any rating. The traditional recommendation systems assume that users are independent and identically distributed and ignore the connections among users whereas the recommendation actually is a social activity. This dissertation aims to enhance NMF-based recommendation systems by utilizing the imputation method and limiting the errors that are introduced in the system. External information such as trust network and item categories are incorporated into NMF-based recommendation systems through the imputation. The proposed approaches impute various subsets of the missing ratings. The subsets are defined based on the total number of the ratings of the user or item before the imputation, such as impute the missing ratings of New-Users, New-Items, or cold-start users or items that suffer from the lack of the ratings. In addition, several factors are analyzed that affect the prediction accuracy when the imputation method is utilized with NMF-based recommendation systems. These factors include the total number of the ratings of the user or item before the imputation, the total number of imputed ratings for each user and item, the average of imputed rating values, and the value of imputed rating values. In addition, several strategies are applied to select the subset of missing ratings for the imputation that lead to increasing the prediction accuracy and limiting the imputation error. Moreover, a comparison is conducted with some popular methods that are in common with the proposed method in utilizing the imputation to handle the lack of ratings, but they differ in the source of the imputed ratings. Experiments on different large-size datasets are conducted to examine the proposed approaches and analyze the effects of the imputation on accuracy. Users and items are divided into three groups based on the total number of the ratings before the imputation is applied and their recommendation accuracy is calculated. The results show that the imputation enhances the recommendation system by capacitating the system to recommend items to New-Users, introduce New-Items to users, and increase the accuracy of the cold-start users and items. However, the analyzed factors play important roles in the recommendation accuracy and limit the error that is introduced from the imputation.