Estimation of the closeness of a topical text to the most rational (i.e., standard) form of expression of the sense corresponding to it in the given natural language requires the existence of some representative reference collection (or corpus) of texts that reflect the most significant concepts of a given topical area and relationships between them. In the current study, the problem of forming such collection is solved by using the occurrence of words from abstracts of articles, selected by an expert, in documents under estimation for inclusion into the reference corpus. The estimation is based on the comparison of values for the 5-th percentile of the empirical distribution corresponding to an array of fractions for nonzero values of the term frequency (TF) for separate phrases within each abstract relative to the analyzed document. Herewith, the most significant documents for the reference corpus will be those with the maximal value of the mentioned percentile. Experiments show that the proposed solution gives at least a fivefold reduction in the number of documents of those that are minimally relevant to a given topical area when implementing selection for the reference corpus.

The article examines the indicator of a complex correlation measure applied to the task of recognizing congestive heart failure by a heart rate signal. It is shown that the features of the Poincaré plot for the sequence of cardiointervals can be evaluated both by statistical descriptors and by special functions reflecting the temporal organization of the analyzed time series. Statistical parameters describe only the spatial distribution of points in the scattering region. The multiple delay correlation function, which includes the complex correlation measure, allows identifying the features of the dynamics of the heart rhythm in pathology, including the presence of nonlinear components. It is the analysis of the structure of the time series that can provide additional information, thereby providing a more complete description of the differences between the classes of norm and pathology. A comparative theoretical and experimental analysis of the studied descriptors was carried out. The experiments were carried out on samples of experimental data of the classes of normal heart rate and heart rate obtained from the databases of the PhysioNet portal. Statistical indicators of the analyzed functions are investigated, a statistical T-test of the Student for the presence of significant differences in these classes is carried out. A method is proposed for recognizing the congestive heart failure in a multidimensional space of consecutive complex correlation measure values obtained for different time delays. It is shown that for such a description a reliable classification of pathology can be achieved against the background of a normal heart rhythm.

Probabilistic models of noisy discrete source coding and object classification are studied. For these models, the appropriate minimal information amounts as the functions of a given admissible error probability are defined and the strictly decreasing lower bounds to these functions are constructed. The defined functions are similar to the rate-distortion function known in the information theory and the lower bounds to the these functions yield a minimal error probability subject to a given value of the processed information amount. So, the obtained bounds are the bifactor fidelity criterions in source coding and object classification tasks.

The paper proposes a weakly supervised binary classification method which combines manifold regularization and fuzzy clustering ensemble methodologies. We assume that the class labels can be fully supervised, defined in terms of a probability distribution or not specified at all. The co-association matrix of fuzzy clustering ensemble is used as the similarity matrix. This matrix is represented in a low-rank form that significantly speeds up calculations and saves memory. Numerical experiments using Monte Carlo modeling demonstrate the efficiency of the method.

Accurate detection of the leaf edge in scanned images is critical for the correct determination of the plant leaves morphometric parameters, which play the role of biological indicators of the ecological situation. Due to the peculiarities of the scanning process, the colored shadow from the leaves appears in the image, which makes it difficult to use standard methods of segmentation and alpha matting. We propose here a new algorithm for segmenting a leaf plate using a parametric representation of the shadow and background in the color space by support vectors and a structure transfer filtering based on a normal-gamma model. The combination of these methods makes it possible to consider both color information and information about sharp changes in image intensity to accurately determine the leaf edges.

Ultrasound image is widely used medical imaging modality for the diagnosis of diseases. In this work, a novel hybrid approach for classification of diseased kidney medical ultrasound images is proposed. The segmented images are passed through the stage of feature extraction. Different features, namely, Haralick, shape, wavelet, Tamura, and histogram oriented gradient features are used for classification. The method focuses on recognizing the most dominant feature set that characterizes the texture of a stone, cyst or normal kidney ultrasound image. Extracted features are individually used for classification by three different classifiers namely k-nearest neighborhood, fuzzy k-nearest neighborhood, and support vector machine. The hybrid approach for feature extraction containing the combination of wavelet and shape features shows the optimal classification accuracy. The efficiency of the method is tested with performance parameters such as accuracy, sensitivity, and specificity. The results obtained show the efficiency of the proposed method.

In this paper, we propose a robust visual tracking algorithm based on local dissimilarity map (LDM) and Kalman filter (KF). Firstly, we model the motion model component of the proposed tracker by using the KF. Then, we apply the LDM into the object matching model to measure the local dissimilarities between the target and the sampled candidates in each frame of the given image sequence. Experimental results on several image sequences illustrate that the proposed method performs well in several challenging aspects of real world scenes.

This article focuses on clustering tasks, normalization of multiple parameters under consideration and applying normalized adaptive league championship algorithm (NALCA) for minimization of makespan of tasks. Minimization of processing time and optimum resource utilization are fundamental requirements of any computing service provider. Scheduling performs crucial role in managing processing capabilities and resource utilization. Clustering is applied so that tasks are divided into various clusters and then scheduling is applied. In order to make comparative analysis of clustering, three clustering algorithms are implemented to have statistical details after applying each clustering algorithm. After experimentation it has been found out that hierarchical clustering with equal opportunity model with normalization LCA has gained outstanding performance enhancements.

Several video coding standards and techniques have been introduced for multimedia applications, particularly h.26x series for video processing. These standards employ motion estimation process for reducing the amount of data that is required to store or transmit the video. Motion estimation process is an inextricable part of the video coding as it removes the temporal redundancy between successive frames of video sequences. This paper is about these motion estimation algorithms, their search procedures, complexity, advantages, and limitations. A survey of motion estimation algorithms including full search algorithm, many fast search, and fast full search block-based algorithms has been presented. An evaluation of up to date motion estimation algorithms, based on a number of empirical results on several test video sequences, is presented as well.

In this paper, we present the main aspects of artificial intelligence application in ophthalmology for the diagnosis and treatment of eye diseases on the example of developing a computer system for personalizing retinal laser photocoagulation. Approaches to the automation of eye disease prediction and treatment based on fundus images are described. Four problems of applying the neural network approach are highlighted. Decision support information technology for personalizing laser treatment of diabetic macular edema and identifying prognostic factors of surgical outcome using methods of intellectual analysis of large unstructured data is described. The system allows the doctor to form a plan of optimal coagulation arrangement for retinal laser coagulation for each case, to predict the quality of laser coagulation depending on the initial data on the localization and severity of edema and to improve his skills by comparing the result of coagulation performed and the coagulation plan proposed by the system.