Nowadays the largest E-commerce or E-service websites offer millions of products for sale. A Recommender system is defined as software used by such websites for recommending commercial or noncommercial product items to users according to the users’ tastes. In this project, we develop a recommender system for a private multimedia web service company. In particular, we devise three recommendation engines using different data filtering methods – named weighted-average, K-nearest neighbors, and item-based – which are based on collaborative filtering techniques, which work by recording user preferences on items and by anticipating the future likes and dislikes of users by comparing the records, for prediction of user preference. To acquire proper input data for the three engines, we retrieve data from database using three data collection techniques: active filtering, passive filtering, and item-based filtering. For experimental purpose we compare prediction accuracy of those three recommendation engines with the results from each engine and additionally we evaluate the performance of weighted-average method using an empirical analysis approach – a methodology which was devised for verification of predictive accuracy.

This work presents improvements to Monte Carlo Localization (MCL) for a mobile robot using computer vision. Solutions to the localization problem aim to provide fine resolution on location approximation, and also be resistant to changes in the environment. One such environment change is the kidnapped/teleported robot problem, where a robot is suddenly transported to a new location and must re-localize. The standard method of "Augmented MCL" uses particle filtering combined with addition of random particles under certain conditions to solve the kidnapped robot problem. This solution is robust, but not always fast. This work combines Histogram of Oriented Gradients (HOG) computer vision with particle filtering to speed up the localization process. The major slowdown in Augmented MCL is the conditional addition of random particles, which depends on the ratio of a short term and long term average of particle weights. This ratio does not change quickly when a robot is kidnapped, leading the robot to believe it is in the wrong location for a period of time. This work replaces this average-based conditional with a comparison of the HOG image directly in front of the robot with a cached version. This resulted in a speedup ranging from from 25.3% to 80.7% (depending on parameters used) in localization time over the baseline Augmented MCL.

<p> Bayesian filtering refers to the process of sequentially estimating the current state of a complex dynamic system from noisy partial measurements using Bayes' rule. This thesis considers Bayesian filtering as applied to an important class of state estimation problems, which is describable by a discrete-time nonlinear state-space model with additive Gaussian noise. It is known that the conditional probability density of the state given the measurement history or simply the posterior density contains all information about the state. For nonlinear systems, the posterior density cannot be described by a finite number of sufficient statistics, and an approximation must be made instead.</p> <p> The approximation of the posterior density is a challenging problem that has engaged many researchers for over four decades. Their work has resulted in a variety of approximate Bayesian filters. Unfortunately, the existing filters suffer from possible divergence, or the curse of dimensionality, or both, and it is doubtful that a single filter exists that would be considered effective for applications ranging from low to high dimensions. The challenge ahead of us therefore is to derive an approximate nonlinear Bayesian filter, which is theoretically motivated, reasonably accurate, and easily extendable to a wide range of applications at a minimal computational cost.</p> <p> In this thesis, a new approximate Bayesian filter is derived for discrete-time nonlinear filtering problems, which is named the cubature Kalman filter. To develop this filter, it is assumed that the predictive density of the joint state-measurement random variable is Gaussian. In this way, the optimal Bayesian filter reduces to the problem of how to compute various multi-dimensional Gaussian-weighted moment integrals. To numerically compute these integrals, a third-degree spherical-radial cubature rule is proposed. This cubature rule entails a set of cubature points scaling linearly with the state-vector dimension. The cubature Kalman filter therefore provides an efficient solution even for high-dimensional nonlinear filtering problems. More remarkably, the cubature Kalman filter is the closest known approximate filter in the sense of completely preserving second-order information due to the maximum entropy principle. For the purpose of mitigating divergence, and improving numerical accuracy in systems where there are apparent computer roundoff difficulties, the cubature Kalman filter is reformulated to propagate the square roots of the error-covariance matrices. The formulation of the (square-root) cubature Kalman filter is validated through three different numerical experiments, namely, tracking a maneuvering ship, supervised training of recurrent neural networks, and model-based signal detection and enhancement. All three experiments clearly indicate that this powerful new filter is superior to other existing nonlinear filters. </p> / Thesis / Doctor of Philosophy (PhD)

No description available.

No description available.

No description available.

The Kalman filter is an important component of optimal estimation theory. It has applications in a wide range of high performance control systems including navigational, fire control, and targeting systems. The Kalman filter, however, has not been utilized to its full potential due to the limitations of its inherent computational intensiveness which requires "off-line" processing or allows only low bandwidth real-time applications. The recent advances in VLSI circuit technology have created the opportunity to design algorithms and data structures for direct implementation in integrated circuits. A systolic architecture is a concept which allows the construction of massively parallel systems in integrated circuits and has been utilized as a means of achieving high data rates. A systolic system consists of a set of interconnected processing elements, each capable of performing some simple operation. The design of a processing element in an orthogonal systolic architecture will be investigated using the state of the art in VLSI technology. The goal is to create a high speed, high precision processing element which is adaptive to a highly configurable systolic architecture. In order to achieve the necessary high computational throughput, the arithmetic unit of the processing element will be implemented using the Logarithmic Number System. The Systolic architecture approach will be used in an attempt to implement a Kalman filtering system with both a high sampling rate and a small package size. The design of such a Kalman filter would enable this filtering technology to be applied to the areas of process control, computer vision, and robotics.

<p> The rapid increasing bandwidth requirement of communication systems demands powerful numerical simulation tools for optics fiber. The computational efficient, memory saving and stable are of the most important characteristics for any simulation tools used for long-haul and broadband optics fiber. An optimized split-step digital filtering method is developed in this paper. The concept of Fourier integral and Fourier series are used in extracting a FIR filter which is used to fit the original transfer function. A further optimization process which employs windowing technique to improve computation efficiency had also been done. Compared with split-step frequency method, our method improves the computation efficiency. Only simple shifts and multiplications are needed in our method. This optimized digital filtering method differs from the former digital filtering method in a sense that the filter length of the FIR filter we extracted is reduced to a very small number. The computation time can be saved as much as 96% than before. This method can also be used to solve coupled nonlinear Schrodinger equation which governs polarization mode dispersion effect in fibers. A new simulation scheme for PMD is proposed to save computation time. The propagation results shows good accordance to those already published results. </p> / Thesis / Master of Applied Science (MASc)

The focus of this research is to develop and extend the application of existing technologies to enhance seismic reservoir characterization. The chapters presented in this dissertation constitute five individual studies consisting of three applications of the Radon transform, one aspect of acoustic wave propagation, and a pilot study of generating a stochastic reservoir model. The first three studies focus on the use of the Radon transform to enhance surface-recorded, controlled-source seismic data. First, the use of this transform was extended to enhance diffraction patterns, which may be indicative of subsurface fractures. The geometry of primary reflections and diffractions on synthetic common-shot-gather data indicate that Radon filters can predict and model primary reflections upon inverse transformation. These modeled primaries can then be adaptively subtracted from the input gather to enhance the diffractions. Second, I examine the amplitude distortions at near and far offsets caused by free-surface multiple removal using Radon filters. These amplitudes are often needlessly reduced due to a truncation effect when the commonly used, unweighted least-squares solution is applied. Synthetic examples indicate that a weighted solution to the transformation minimizes this effect and preserves the reflection amplitudes. Third, a novel processing flow was developed to generate a stacked seismic section using the Radon transform. This procedure has the advantage over traditional summation of normal moveout corrected common midpoint gathers because it circumvents the need to perform manual and interpretive velocity analysis. The fourth study involves the detection of thin layers in periodic layerstacks. Numerical modeling of acoustic wave propagation suggests that the sinusoidal components of an incident signal with a wavelength that corresponds to the periodicity of the material be preferentially reflected. Isolating the different portions of the reflected wavefield and calculating the energy spectra may provide evidence of thin periodic layers which are deterministically unresolvable on their own. Object-based reservoir modeling often incorporates the use of lithology logs, deterministic seismic interpretation, architectural element analysis, geologic intuition, and modern and outcrop analogs. This last project consists of a pilot study where a more quantitative approach to define the statistical parameters currently derived through geologic intuition and analogs was developed. This approach utilizes a simulated annealing optimization technique for inversion and the pilot study shows that it can improve the correlation between synthesized and control logs. / Ph. D.

Speech coding or speech compression is one of the important aspects of speech communications nowadays. By coding the speech, the speed needed to transmit the digitized speech, called the bit rate, can be reduced. This means that for a certain speech communications channel, the lower the bit rate of the speech coding, the more communicating parties can be carried on that channel. This research has as its main application the extraction of the parameters of human speech for speech coding purposes. We propose an RLS-based cascade adaptive filter structure that can significantly reduce the computational effort required by the RLS algorithm for inverse filtering types of applications. We named it the Cascade RLS with Subsection Adaptation (CRLS-SA) algorithm. The reduction in computational effort comes from the fact that, for inverse filtering applications, the gradients of each section in the cascade are almost uncorrelated with the gradients in other sections. Hence, the gradient autocorrelation matrix is assumed to be block diagonal. Since we use a second order filter for each section, the computation of the adaptation involves only the 2x2- gradient autocorrelation matrix for that section, while still being based on a global minimization criterion. The gradient signal of a section itself is defined as the derivative of the overall output error with respect to the coefficients of the particular section, which can be computed efficiently by passing the overall output of the cascade to a filter with coefficients that are derived from the coefficients of that section. The computational effort of the CRLS-SA algorithm is approximately 20*L*N/2, where L is the data record length and N is the order of the filter. We analyze the convergence rate of the CRLS-SA algorithm based on the convergence time constant concept, which is the ratio of the condition number and the sensitivity. The CRLS- SA structure is shown to satisfy the DeBrunner-Beex conjecture which says that a structure with a smaller convergence time constant converges faster than a structure with a larger convergence time constant. We show that CRLS-SA converges faster than the Direct Form RLS (DFRLS) algorithm and that its convergence time constant is lower than that of the direct form. The convergence behavior is verified by looking at how fast the estimated system approaches the true system. Here we use the Itakura distance as the measure of closeness between the estimated and the true system. We show that the Itakura distance associated with the CRLS-SA algorithm approaches zero faster than that associated with the direct form RLS algorithm. The CRLS-SA algorithm is applied in this dissertation to general linear prediction, to the direct adaptive computation of the LSF and their representation in quantized form using a split vector quantization (VQ) approach, and to the detection and tracking of the frequencies in signals consisting of multiple sinusoids in noise. / Ph. D.