In this paper, a fusion of K models of full-rank weighted nonnegative tensor factor two-dimensional deconvolution (K-wNTF2D) is proposed to separate the acoustic sources that have been mixed in an underdetermined reverberant environment. The model is adapted in an unsupervised manner under the hybrid framework of the generalized expectation maximization and multiplicative update algorithms. The derivation of the algorithm and the development of proposed full-rank K-wNTF2D will be shown. The algorithm also encodes a set of variable sparsity parameters derived from Gibbs distribution into the K-wNTF2D model. This optimizes each sub-model in K-wNTF2D with the required sparsity to model the time-varying variances of the sources in the spectrogram. In addition, an initialization method is proposed to initialize the parameters in the K-wNTF2D. Experimental results on the underdetermined reverberant mixing environment have shown that the proposed algorithm is effective at separating the mixture with an average signal-to-distortion ratio of 3 dB.
The process of accurate localization of the basic components of human faces (i.e., eyebrows, eyes, nose, mouth, etc.) from images is an important step in face processing techniques like face tracking, facial expression recognition or face recognition. However, it is a challenging task due to the variations in scale, orientation, pose, facial expressions, partial occlusions and lighting conditions. In the current paper, a scheme includes the method of three-hierarchal stages for facial components extraction is presented; it works regardless of illumination variance. Adaptive linear contrast enhancement methods like gamma correction and contrast stretching are used to simulate the variance in light condition among images. As testing material
... Show MoreThe primary objective of this paper is to improve a biometric authentication and classification model using the ear as a distinct part of the face since it is unchanged with time and unaffected by facial expressions. The proposed model is a new scenario for enhancing ear recognition accuracy via modifying the AdaBoost algorithm to optimize adaptive learning. To overcome the limitation of image illumination, occlusion, and problems of image registration, the Scale-invariant feature transform technique was used to extract features. Various consecutive phases were used to improve classification accuracy. These phases are image acquisition, preprocessing, filtering, smoothing, and feature extraction. To assess the proposed
... Show MoreThe aim of this paper is to present a method for solving third order ordinary differential equations with two point boundary condition , we propose two-point osculatory interpolation to construct polynomial solution. The original problem is concerned using two-points osculatory interpolation with the fit equal numbers of derivatives at the end points of an interval [0 , 1] . Also, many examples are presented to demonstrate the applicability, accuracy and efficiency of the method by compared with conventional method .
In this paper, the method of estimating the variation of Zenith Path Delay (ZPD) estimation method will be illustrate and evaluate using Real Time Kinematic Differential Global Positioning System (RTK-DGPS). The GPS provides a relative method to remotely sense atmospheric water vapor in any weather condition. The GPS signal delay in the atmosphere can be expressed as ZPD. In order to evaluate the results, four points had been chosen in the university of Baghdad campus to be rover ones, with a fixed Base point. For each rover position a 155 day of coordinates measurements was collected to overcome the results. Many models and mathematic calculations were used to extract the ZPD using the Matlab environment. The result shows that the ZPD valu
... Show MoreThe main aim of this paper is to study how the different estimators of the two unknown parameters (shape and scale parameter) of a generalized exponential distribution behave for different sample sizes and for different parameter values. In particular,
. Maximum Likelihood, Percentile and Ordinary Least Square estimators had been implemented for different sample sizes (small, medium, and large) and assumed several contrasts initial values for the two parameters. Two indicators of performance Mean Square Error and Mean Percentile Error were used and the comparisons were carried out between different methods of estimation by using monte carlo simulation technique .. It was obse
... Show MoreAspect-based sentiment analysis is the most important research topic conducted to extract and categorize aspect-terms from online reviews. Recent efforts have shown that topic modelling is vigorously used for this task. In this paper, we integrated word embedding into collapsed Gibbs sampling in Latent Dirichlet Allocation (LDA). Specifically, the conditional distribution in the topic model is improved using the word embedding model that was trained against (customer review) training dataset. Semantic similarity (cosine measure) was leveraged to distribute the aspect-terms to their related aspect-category cognitively. The experiment was conducted to extract and categorize the aspect terms from SemEval 2014 dataset.
A mixture of algae biomass (Chrysophyta, Cyanophyta, and Chlorophyte) has been investigated for its possible adsorption removal of cationic dyes (methylene blue, MB). Effect of pH (1-8), biosorbent dosage (0.2-2 g/100ml), agitated speed (100-300), particle size (1304-89μm), temperature (20-40˚C), initial dye concentration (20-300 mg/L), and sorption–desorption were investigated to assess the algal-dye sorption mechanism. Different pre-treatments, alkali, protonation, and CaCl2 have been experienced in order to enhance the adsorption capacity as well as the stability of the algal biomass. Equilibrium isotherm data were analyzed using Langmuir, Freundlich, and Temkin models. The maximum dye-sorption capacity was 26.65 mg/g at pH= 5, 25
... Show MoreThis paper is concerned with introducing and studying the o-space by using out degree system (resp. i-space by using in degree system) which are the core concept in this paper. In addition, the m-lower approximations, the m-upper approximations and ospace and i-space. Furthermore, we introduce near supraopen (near supraclosed) d. g.'s. Finally, the supra-lower approximation, supraupper approximation, supra-accuracy are defined and some of its properties are investigated.