Sharif Digital Repository

Hidden Markov Model (HMM)-based synthesis (HTS) has recently been confirmed to be the most effective method in generating natural speech. However, it lacks adequate context generalization when the training data is limited. As a solution, current study provides a new context-dependent speech modeling framework based on the Gaussian Conditional Random Field (GCRF) theory. By applying this model, an innovative speech synthesis system has been developed which can be viewed as an extension of Context-Dependent Hidden Semi Markov Model (CD-HSMM). A novel Viterbi decoder along with a stochastic gradient ascent algorithm was applied to train model parameters. Also, a fast and efficient parameter... 

In this letter, we study the joint source-channel coding problem of communicating the linear functions of correlated Gaussian sources over a Gaussian multiple access channel. In this setup, each transmitter observes one of two correlated sources. The receiver aims to reconstruct a linear combination of both correlated Gaussian sources within an average distortion level. For this setup, we first obtain a new lower bound on the achievable distortion and then, we propose a novel lattice-based hybrid transmission scheme and its achievable distortion is derived. We show that our proposed scheme can achieve the optimal distortion under some conditions  

In this paper a sub-optimal control technique is proposed for a linear Gaussian system with a few distributed interacting sub-systems. Controller of each sub-system has only access to its own measurement and a noisy version of measurement and control vectors of other sub-systems that are communicated through an Additive White Gaussian Noise (AWGN) communication network. The power to be allocated to each transmitter antenna is calculated so that the received signal is the transmitted signal plus additive white Gaussian noise. Under some conditions it is shown that the proposed sub-optimal control technique results in bounded mean square stability. The satisfactory performance of the proposed... 

Wide-waisted Gaussian beam (GB) expansion and point matching methods are discussed in this study as promising techniques for the expansion of electromagnetic waves into GBs. These methods combined with the GB tracing method provide an appropriate scheme for wave propagation analysis in large-scale problems. Details of the wide-waisted GB expansion and point matching methods are provided, and their accuracy and validity regions are discussed. A combination of the aforementioned expansion and tracing methods is utilised to analyse a simple outdoor high-frequency wave propagation problem. The results are compared to the results of the well-known ray tracing method, showing good agreement  

We have developed a generalized closed-form representation of the virtually imaged phased array (VIPA) output profile based on a 3D vectorial Gaussian beam formulation. Tracing the input vectorial elliptic Gaussian beam (GB), the output profile of the VIPA is represented as an array of virtual images. Furthermore, employing GB propagation and summing up all the virtual image profiles, the VIPA output profile has been calculated on a vertical plane at an arbitrary distance from the device. Finally, the derived closed-form representation is applied to the different sources (i.e., circular and elliptic GBs) and the results are compared with previous work. © 2009 Optical Society of America  

This paper is a survey of semi-blind source separation approaches. Since Gaussian iid signals are not separable, simplest priors suggest to assume non Gaussian iid signals, or Gaussian non iid signals. Other priors can also been used, for instance discrete or bounded sources, positivity, etc. Although providing a generic framework for semi-blind source separation, Sparse Component Analysis and Bayesian ICA will just sketched in this paper, since two other survey papers develop in depth these approaches. © 2006 i6doc.com publication. All rights reserved  

The non-paraxial and vectorial effects have an important role in the dynamics of highly confined beam in nonlinear medium. In this paper, we study the vectorial non-paraxial propagation of Gaussian field in the saturable absorber media numerically with the aim of nano-lithography. It is shown that in the optimal regimes for nanolithography, the mutual coupling between components of optical field and mode coupling between different orders of Hermite-Gaussian mode lead to generation of various patterns upon propagation through saturable absorber media. The vectorial effect is responsible for the symmetry breaking. For converging input beam, while the intense part of the beam goes toward the... 

An efficient method to automatically recognize basic human actions is proposed to improve the communication between a human and a computer. Human actions are considered as patterns generated by complex non-linear dynamical models. A non-linear dynamical model is used to represent human actions. Gaussian process dynamical models are used to capture the spatial and temporal behaviors of actions. To make the process more efficient a 7-dimensional feature is extracted for each action. Although the extracted feature vector is compact compared to a high-dimensional temporal pattern, it can efficiently discriminate among different actions. The tests run on CMU MoCap database with SVM show promising... 

The Ensemble Kalman Filter (EnKF) is a Monte Carlo based method to assimilate the measurement data sequentially in time. Although, EnKF has some advantages over the other Kalman based methods to deal with non-linear and/or high dimensional reservoir models, it also suffers from deficiency in estimation of non-Gaussian parameters. In this work, we propose a re-parameterization method to handle non-Gaussian parameters via Ensemble Kalman Filter framework. For this purpose, concept of cumulative distribution function transformation has been used. In addition, the statistics of prior information have been aggregated in the state vector in order to capture the prior uncertainties of non-Gaussian... 

In this paper we address the problem of image interpolation using Gaussian Mixture Models (GMM) as a prior. Previous methods of image restoration with GMM have not considered spatial (geometric) distance between patches in clustering, failing to fully exploit the coherency of nearby patches. The GMM framework in our method for image interpolation is based on the assumption that the accumulation of similar patches in a neighborhood are derived from a multivariate Gaussian probability distribution with a specific covariance and mean. An Expectation Maximization-like (EM-like) algorithm is used in order to determine patches in a cluster and restore them. The results show that our image... 

We consider the Gaussian Dirty Tape Channel (DTC) Y=X+S+Z, where S is an additive Gaussian interference known causally to the transmitter. The general expression maxPU, f(.), X=f(U,S) I(U;Y) is proposed for the capacity of this channel. For linear assignment to f(.), i.e., X=U-βS, this expression leads to the compensation strategy proposed previously by Willems to obtain an achievable rate for the DTC. We show that linear assignment to f(.) is optimal under the condition that there exists a real number β* such that the pair (X+β*S, U) is independent of the interference S. Furthermore, by applying a time-sharing technique to the achievable rate derived by linear assignment to f (.), an... 

We analyze the VIPA demultiplexer for different reflection profiles, i.e., uniform, graded, and Gaussian. It is shown that VIPA demultiplexer characteristics can be improved dramatically by innovative reflection profile engineering. These improvements enhance the performance of demultiplexers in UWB-Over-Fiber systems  

We have developed a Gaussian Process Regression method with adaptive kernels for concealment of the missing macro-blocks of block-based video compression schemes in a packet video system. Despite promising results, the proposed algorithm introduces a solid framework for further improvements. In this paper, the problem of estimating lost macro-blocks will be solved by estimating the proper covariance function of the Gaussian process defined over a region around the missing macro-blocks (i.e. its kernel function). In order to preserve block edges, the kernel is constructed adaptively by using the local edge related information. Moreover, we can achieve more improvements by local estimation of... 

The discrete memoryless state-dependent relay channel (SD-RC) is considered in this study. Two main cases are investigated: SD-RC with non-causal channel state information (CSI) and SD-RC with causal CSI. In each case, the SD-RC with partial CSI at the source and the relay is considered. As special cases it includes three different situations in which perfect CSI is available: (i) only at the source, (ii) only at the relay and (iii) both at the source and the relay. For the non-causal situation, the authors establish lower bound on capacity (achievable rate) of the SD-RC, using Gel'fand-Pinsker coding at the nodes informed of CSI and compress-and-forward (CF) strategy at the relay. Using the... 

In this letter, we consider two problems of sending a bivariate Gaussian source through a two-user Gaussian non-orthogonal multiple-access channel. We provide a new distortion outer bound for the considered joint source-channel coding problems, which significantly reduces the gap between inner and outer bounds of the distortion for correlation coefficients below a threshold. Furthermore, numerical results show, that under some conditions, our proposed outer bound matches with the inner bound obtained via the separate source-channel coding scheme, which proves the optimality of this scheme. © 1997-2012 IEEE  

It is of fundamental importance to find algorithms obtaining optimal performance for learning of statistical models in distributed and communication limited systems. Aiming at characterizing the optimal strategies, we consider learning of Gaussian Processes (GP) in distributed systems as a pivotal example. We first address a very basic problem: how many bits are required to estimate the inner-products of some Gaussian vectors across distributed machines? Using information theoretic bounds, we obtain an optimal solution for the problem which is based on vector quantization. Two suboptimal and more practical schemes are also presented as substitutes for the vector quantization scheme. In... 

We report on a study of the Tehran Price Index (TEPIX) from 2001 to 2006 as an emerging market that has been affected by several political crises during the recent years, and analyze the non-Gaussian probability density function (PDF) of the log returns of the stock prices. We show that while the average of the index did not fall very much over the time period of the study, its day-to-day fluctuations strongly increased due to the crises. Using an approach based on multiplicative processes with a detrending procedure, we study the scale-dependence of the non-Gaussian PDFs, and show that the temporal dependence of their tails indicates a gradual and systematic increase in the probabilityof... 

In this paper a new method for out-of-focus blur estimation and restoration in presence of Gaussian noise is proposed. This method is based on Bispectrum transformation. It has been proved that functions with Gaussian distribution are suppressed in Bispectrum domain. The image blur criterion is estimated using information in image Bispectrum transformation. Then, an inverse filter is used for blurred image restoration. Experimental results show satisfactory performance of presented method in degradation function parameter estimation in comparison with other algorithms. This superiority is especially obvious in restoration of noisy images, since most of the methods do not take into account... 

In this letter, a new adaptive Gaussian process (GP) frame work for signal extrapolation is proposed. Signal extrapolation is an essential task in many applications such as concealment of corrupted data in image and video communications. While possessing many interesting properties, Gaussian process priors with inappropriate stationary kernels may create extremely blurred edges in concealed areas of the image. To address this problem, we propose adaptive non-stationary kernels in a Gaussian process framework. The proposed adaptive kernel functions are defined based on the hypothesized edges of the missing areas. Experimental results verify the effectiveness of the proposed method compared to... 

In this paper, we address the problem of recovering degraded images using multivariate Gaussian mixture model (GMM) as a prior. The GMM framework in our method for image restoration is based on the assumption that the accumulation of similar patches in a neighborhood are derived from a multivariate Gaussian probability distribution with a specific covariance and mean. Previous methods of image restoration with GMM have not considered spatial (geometric) distance between patches in clustering. Our conducted experiments show that in the case of constraining Gaussian estimates into a finite-sized windows, the patch clusters are more likely to be derived from the estimated multivariate Gaussian...