Search

In this paper we investigate the proprieties of automatic speaker recognition (ASR) to develop a system for voice pathologies detection, where the model does not correspond to a speaker but it corresponds to group of patients who shares the same diagnostic. One of essential part in this topic is the database (described later), the samples voices (healthy and pathological) are chosen from a German database which contains many diseases, spasmodic dysphonia is proposed for this study. This problematic can be solved by statistical pattern recognition techniques where we have proposed the mel frequency cepstral coefficients (MFCC) to be modeled first, with gaussian mixture model (GMM) massively used in ASR then, they are modeled with support vector machine (SVM). The obtained results are compared in order to evaluate the more preferment classifier. The performance of each method is evaluated in a term of the accuracy, sensitivity, specificity. The best performance is obtained with 12 coefficientsMFCC, energy and second derivate along SVM with a polynomial kernel function, the classification rate is 90% for normal class and 93% for pathological class.This work is developed under MATLAB

Background Emotions play a central role in mental disorder and especially in depression. They are sensed in the body, and it has recently been shown in healthy participants that these sensations can be differentiated between emotions. The aim of the current study was to assess bodily sensations for basic emotions induced by emotion eliciting pictures in depression. Methods 30 healthy controls (HC), 30 individuals suffering from Major depressive disorder (MDD) without medication use (MDDnm) and 30 individuals with MDD with medication use (MDDm) were shown emotional and neutral pictures and were asked to paint areas in an empty body silhouette where they felt an increase or decrease in activation. Body sensation maps were then calculated and statistical pattern recognition applied. Results Results indicated statistically separable activation patterns for all three groups. MDDnm showed less overall activation than HCs, especially in sadness and fear, while MDDm showed stronger deactivation for all emotions than the other two groups. Conclusions We could show that emotion experience was associated with bodily sensations that are weaker in depression than in healthy controls and that antidepressant medication was correlated with an increased feeling of bodily deactivation. Results give insights into the relevance for clinicians to acknowledge bodily sensations in the treatment of depression.

This study is devoted to the statistical method of classification based on nonlinear regression. The approaches used to implement it in solving the recognition problem of printed and handwritten characters are presented. Its implementation in assessing the health of the systems of the human body according to the parameters of peripheral blood is presented for the first time. The optimal structure of the polynomials is proposed. The properties of the probability estimates generated by the method are described. The structure of the sets used to train it is analyzed.

Recently, Ecuador has shown low productivity; for this reason it was necessary to do an exhaustive analyzes about technogogical innovation and its influence in the economical development. For this researching it was important to get information of National and International Organizations that are in relationship with Reasearching, Development and Innovation. The collection andanalyzes of data was executed with technological programs such as: EXCEL and SPSS. Thisinformation was proved through hypothesis, using the statistical pattern “chi-cuadrado” (X^2).This kind of results compares the verified ditribution with the planned distribution. So, it has been possible to get reliable results. These results are based on the differences that get the probability distribution to contrast. With this kind of analyzes it was proved that technological innovation not contributed meaningfully in the economical developement. Furthermore, proposal has beben submitted of public politics and technological innovation to help the productivity and to get an excellent knowledge based on the intellect and be possible to believe, value and share to be used in an efficient way.

Structural health monitoring methods are broadly classified into two categories: data-driven methods via statistical pattern recognition and physics-based methods through finite elementmodel updating. Data-driven structural health monitoring faces the challenge of data insufficiency that renders the learned model limited in identifying damage scenarios that are not contained in the training data. Model-based methods are susceptible to modeling error due to model idealizations and simplifications that make the finite element model updating results deviate from the truth. This study attempts to combine the merits of data-driven and physics-based structural health monitoring methods via physics-guided machine learning, expecting that the damage identification performance can be improved. Physics-guided machine learning uses observed feature data with correct labels as well as the physical model output of unlabeled instances. In this study, physics-guided machine learning is realized with a physics-guided neural network. The original modal-property based features are extended with the damage identification result of finite element model updating. A physics-based loss function is designed to evaluate the discrepancy between the neural network model output and that of finite element model updating. With the guidance from the scientific knowledge contained in finite element model updating, the learned neural network model has the potential to improve the generality and scientific consistency of the damage detection results. The proposed methodology is validated by a numerical case study on a steel pedestrian bridge model and an experimental study on a three-story building model.

Traditional statistical pattern identification methods, such as artificial neural network and support vector machine, have limited ability to identify minor damage of bridges. Deep learning can mine the inherent law and representation level of sample data. As a typical algorithm of deep learning, convolutional neural network is a feedforward neural network with deep structure and convolution calculation, and its ability of image identification is very outstanding. The recurrence graph of structural response can reveal the internal structure, similarity, and damage information. The original structure response signal involves the coupling vibration of vehicle and bridge is filtered and reconstructed by wavelet packet, and then the recurrence graph of different damage cases is obtained, which is used as the input image of convolutional neural network as a new type of damage feature; thus, a damage identification method based on convolutional neural network and recurrence graph is established. The results of numerical simulation and model experiment show that the recurrence graph contains more damage information; compared with the traditional statistical pattern identification methods, convolutional neural network can achieve more accurate feature extraction and identification through intelligent learning layer by layer, so as to realize more accurate identification of damage location and damage degree.

Because hyperspectral remote sensing images have high band dimension and limited training samples, it is hard to achieve acceptable classification results using conventional statistical pattern recognition methods. Nevertheless, support vector machine (SVM) is suitable for hyperspectral image classification owing to its good generalization ability and minimal structural risk. Aimed to address the problems of large subjectivity, excessive time consumption, and low optimization accuracy in traditional SVM parameter selection, we applied genetic algorithm to optimize the model parameters. Using EO-1 Hyperion data as an example, an SVM object classification algorithm was proposed based on genetic algorithm optimization. To determine the effectiveness and superiority of this algorithm, a comprehensive evaluation and comparative analysis were performed with the classification effects of the cross-validation and maximum likelihood methods, respectively. Experimental results demonstrated that the proposed algorithm can effectively avert the aimlessness of artificial parameter selection and automatically optimize the SVM model parameters. It achieved relatively high overall classification accuracy (91.23%), which was 16.42% and 4.48% higher than the maximum likelihood method and cross-validation method, respectively.

Damage localization of damaged structures is an important issue in structural health monitoring. In data-based methods based on statistical pattern recognition, it is necessary to extract meaningful features from measured vibration signals and utilize a reliable statistical technique for locating damage. One of the challenging issues is to extract reliable features from non-stationary vibration signals caused by ambient excitation sources. This article proposes a new energy-based method by using ensemble empirical mode decomposition and Mahalanobis-squared distance to obtain energy-based multivariate features and locate structural damage under ambient vibration and non-stationary signals. The main components of the proposed method include extracting intrinsic mode functions of vibration signals by ensemble empirical mode decomposition, choosing adequate and optimal intrinsic mode functions, partitioning the selected intrinsic mode functions at each sensor into segments with the same dimensions, calculating the intrinsic mode function energy at each segment, preparing energy-based multivariate features at each sensor, computing Mahalanobis-squared distance values, and obtaining a vector of average Mahalanobis-squared distance quantities of all sensors. The major contributions of the proposed method consist of proposing an innovative non-parametric strategy for feature extraction, presenting generalized Pearson correlation function for the selection of optimal intrinsic mode functions, using a simple and effective segmentation algorithm, and applying energy-based features to the process of damage localization. The main advantage of the proposed method is its great applicability to locating single and multiple damage cases. The measured vibration responses of the well-known IASC-ASCE structure are applied to verify the effectiveness and reliability of the proposed energy-based method along with several comparative studies. Results will demonstrate that this approach is highly capable of locating damage under stationary and non-stationary vibration signals attributable to ambient excitations.

This article critically assesses the utility of both personal blog and mass media (on-line newspaper)’s coverage of future technology in forecasting the prospect of industrial technology. By analyzing the statistical pattern of the South Korean media salience in 13 novel industrial technologies in 2015, the study argues that the mass media is more biased than aggregated blogs in depicting promising new technology. In doing so, the authors present a methodical pathway to address the bias with acceleration and skewness index. The article also applies semantic network analysis for the collected textual data of blogs to represent and interpret people's perspective of industrial technologies. By extracting key concepts from PBS index and merging semantic networks of the 13 technologies, the study derives a key insight that nurturing expertise in the coming era of artificial intelligence and robot would become crucial to integrate key technological competence.

Cellular phenotypes emerge from a hierarchy of molecular interactions: proteins interact to form complexes, pathways, and phenotypes. We show that hierarchical networks of protein interactions can be extracted from the statistical pattern of proteome variation as measured across thousands of bacteria and that these hierarchies reflect the emergence of complex bacterial phenotypes. We describe the mathematics underlying our statistical approach and validate our results through gene-set enrichment analysis and comparison to existing experimentally-derived hierarchical databases. We demonstrate the biological utility of our unbiased hierarchical models by creating a model of motility in Pseudomonas aeruginosa and using it to discover a previously unappreciated genetic effector of twitch-based motility. Overall, our approach, SCALES (Spectral Correlation Analysis of Layered Evolutionary Signals), predicts hierarchies of protein interaction networks describing emergent biological function using only the statistical pattern of bacterial proteome variation.

In this work we investigate advanced stochastic methods for solving a specific multidimensional problem related to neural networks. Monte Carlo and quasi-Monte Carlo techniques have been developed over many years in a range of different fields, but have only recently been applied to the problems in neural networks. As well as providing a consistent framework for statistical pattern recognition, the stochastic approach offers a number of practical advantages including a solution to the problem for higher dimensions. For the first time multidimensional integrals up to 100 dimensions related to this area will be discussed in our numerical study.

Adults from Western cultures attribute others’ behavior to personal causes more readily than situational causes; however, little research has explored the developmental origins of this attributional bias. Research has shown that children can use both the statistical patterns present in observed behavior, as well as the verbal framing of the behaviors, to infer personal causes. However, research has not explored whether children also use these factors to infer situational causes. The present study examined the impacts of statistical patterns and verbal framing on four- and six-year-old children’s (n = 218) attributions to personal and situational causes for behavior, as assessed by their explanations for characters’ interactions with toys. In a factorial design the statistical pattern of characters’ behaviors suggested either a personal or situational cause (or neither), and the experimenter’s verbal framing of the behaviors suggested either a personal or situational cause (or neither). Across age groups, children showed a bias toward providing personal explanations. Both statistical pattern and verbal framing influenced causal attributions, but both impacts were asymmetric such that situational cues increased situational explanations relative to neutral cues, but there was no difference in children’s explanations following personal and neutral cues. These results suggest that verbal framing and statistical patterns impact children’s developing social causal attributions, specifically with respect to situational causes, and also that a personal attribution bias emerges early in development.

Structural health monitoring (SHM) is often approached from a statistical pattern recognition or machine learning perspective with the aim of inferring the health state of a structure using data derived from a network of sensors placed upon it. In this paper, two SHM sensor placement optimisation (SPO) strategies that offer robustness to environmental effects are developed and evaluated. The two strategies both involve constructing an objective function (OF) based upon an established damage classification technique and an optimisation of sensor locations using a genetic algorithm (GA). The key difference between the two strategies explored here is in whether any sources of benign variation are deemed to be observable or not. The relative performances of both strategies are demonstrated using experimental data gathered from a glider wing tested in an environmental chamber, with the structure tested in different health states across a series of controlled temperatures.

Structural health monitoring of civil structures is a great necessity for every society due to the key importance of health monitoring of valuable and critical civil structures in order to prevent economic and human losses caused by damage occurrence. This chapter of the book introduces the background and motivation of this process, the main levels and methods of structural health monitoring. Due to advances in sensing and data acquisition systems, data-driven methods have become increasingly popular among civil engineers and researchers owing to simplicity, robustness, and computational efficiency. Data-driven methods are generally based on statistical pattern recognition paradigm. Therefore, the main steps of this paradigm and the concepts of machine learning are clarified in this chapter. One of the major challenging issues in structural health monitoring is environmental and operational variability conditions. Therefore, a part of this chapter is devoted to these conditions and their negative influences. Finally, the aim and scope of this book are highlighted at the end of this chapter.

This chapter provides insight on pattern recognition by illustrating various approaches and frameworks which aid in the prognostic reasoning facilitated by feature selection and feature extraction. The chapter focuses on analyzing syntactical and statistical approaches of pattern recognition. Typically, a large set of features have an impact on the performance of the predictive model. Hence, there is a need to eliminate redundant and noisy pieces of data before developing any predictive model. The selection of features is independent of any machine learning algorithms. The content-rich information obtained after the elimination of noisy patterns such as stop words and missing values is then used for further prediction. The refinement and extraction of relevant features yields in performance enhancements of future prediction and analysis.

In the automatic detection of English pronunciation errors, it is easy to be affected by multiple stress problems, resulting in large errors. Therefore, an automatic detection method of English pronunciation errors based on statistical pattern recognition is proposed. In order to solve the problem of noise in the original speech information, the speech signal is preprocessed by adding windows and framing. Statistical pattern recognition technology is used to recognize the acoustic features of speech signals, and the acoustic model is established. At the same time, the phoneme mispronunciation network is constructed based on the wrong pronunciation features. The acoustic model is used to detect English pronunciation errors automatically under the network of phoneme errors. The experimental results show that the Sentence Rhythm detected by the proposed method is basically consistent with the actual pronunciation, and the actual detection level of the method is improved.

The chemical composition of bee pollens differs greatly and depends primarily on the botanical origin of the product. Therefore, it is a crucially important task to discriminate pollens of different plant species. In our work, we aim to determine the applicability of microscopic pollen analysis, spectral colour measurement, sensory, NIR spectroscopy, e-nose and e-tongue methods for the classification of bee pollen of five different botanical origins. Chemometric methods (PCA, LDA) were used to classify bee pollen loads by analysing the statistical pattern of the samples and to determine the independent and combined effects of the above-mentioned methods. The results of the microscopic analysis identified 100% of sunflower, red clover, rapeseed and two polyfloral pollens mainly containing lakeshore bulrush and spiny plumeless thistle. The colour profiles of the samples were different for the five different samples. E-nose and NIR provided 100% classification accuracy, while e-tongue &gt, 94% classification accuracy for the botanical origin identification using LDA. Partial least square regression (PLS) results built to regress on the sensory and spectral colour attributes using the fused data of NIR spectroscopy, e-nose and e-tongue showed higher than 0.8 R2 during the validation except for one attribute, which was much higher compared to the independent models built for instruments.

Many diseases have no visual cues in the early stages, eluding image-based detection. Today, osteoarthritis (OA) is detected after bone damage has occurred, at an irreversible stage of the disease. Currently no reliable method exists for OA detection at a reversible stage. We present an approach that enables sensitive OA detection in presymptomatic individuals. Our approach combines optimal mass transport theory with statistical pattern recognition. Eighty-six healthy individuals were selected from the Osteoarthritis Initiative, with no symptoms or visual signs of disease on imaging. On 3-y follow-up, a subset of these individuals had progressed to symptomatic OA. We trained a classifier to differentiate progressors and nonprogressors on baseline cartilage texture maps, which achieved a robust test accuracy of 78% in detecting future symptomatic OA progression 3 y prior to symptoms. This work demonstrates that OA detection may be possible at a potentially reversible stage. A key contribution of our work is direct visualization of the cartilage phenotype defining predictive ability as our technique is generative. We observe early biochemical patterns of fissuring in cartilage that define future onset of OA. In the future, coupling presymptomatic OA detection with emergent clinical therapies could modify the outcome of a disease that costs the United States healthcare system $16.5 billion annually. Furthermore, our technique is broadly applicable to earlier image-based detection of many diseases currently diagnosed at advanced stages today.

Understanding how assemblages of invertebrates change over continuous elevational gradients not only generates an understanding of current rules of community assembly but may also be useful for predicting the future distributions of species under global change. Temperature decreases predictably with increasing elevation and, accordingly, gradients in elevation permit the study of adjacent climates within small geographical areas. The present study examines if and how assemblages of moths change with increasing elevation in the eastern French Pyrenees. Elevation had a strong effect on the assemblage composition of moth species in both seasons. The species sets which contributed most to this strong pattern differed completely across seasons. Analysis of restrictions and fidelity to particular elevational ranges generated a set of indicator species which can be used to monitor future changes in distribution. Twelve species were identified as elevation-specific indicators (the ‘predictor set’) from the spring samples and summer samples. We note the strong contrasts between species that produce overall statistical pattern and those that show strong fidelity to particular elevations and discuss this in terms of the biologies of the species concerned. We discuss best practice for the identification and use of indicator species for monitoring future responses to climate change.

Many problems in the analysis of natural terrain surface shapes and the construction of terrain maps to model them remain unsolved. Almost the whole process of thematic interpretation of aerospace information consists of a step-by-step grouping and further data conversion for the purpose of creating a completely definite, problematically oriented picture of the earth’s surface. In this article, we present application of a new method of drawing 3D visibility networks for pattern recognition and its application on terrain surfaces. For the determination of complexity of 3D surface terrain, we use fractal geometry method. We use algorithm for constructing the visibility network to analyse the topological property of networks used in complex terrain surfaces. Terrain models give a fast overview of a landscape and are often fascinating and overwhelmingly beautiful works by artists who invest all their interest and an immense amount of work and know-how, combined with a developed sense of the portrayed landscape, in creating them. At the end, we present modelling of terrain surfaces with topological properties of the visibility network in 3D space.

Feature extraction by time-series analysis and decision making through distance-based methods are powerful and efficient statistical pattern recognition techniques for data-driven structural health monitoring. The motivation of this article is to propose an innovative residual-based feature extraction approach based on AutoRegressive modeling and a novel statistical distance method named as Partition-based Kullback–Leibler Divergence for damage detection and localization by using randomly high-dimensional damage-sensitive features under environmental and operational variability. The key novel element of the proposed feature extraction approach is to establish a two-stage offline and online learning algorithms for extracting the residuals of AutoRegressive model as the main damage-sensitive features. This technique brings the great benefit of reducing the computational time and storage space for feature extraction in long-term monitoring conditions. The major contribution of Partition-based Kullback–Leibler Divergence method is to exploit a partitioning strategy for dividing random features into individual partitions and utilize numerical information of partitioning in distance calculation rather than directly applying random samples. Dealing with the major challenging issue of using the high-dimensional features in decision making and applicability to both correlated and uncorrelated random datasets are the main advantages of Partition-based Kullback–Leibler Divergence method. The accuracy and reliability of the proposed approaches are experimentally validated by two well-known benchmark structures. The stationarity and linearity of measured vibration responses for using in AutoRegressive modeling are evaluated by two hypothesis tests. Comparative studies are also conducted to demonstrate the superiority of the proposed methods over some exciting state-of-the-art techniques. Results show that the methods presented here succeed in detecting and locating damage and make time-saving and efficient tools for feature extraction and damage diagnosis.

As a futuristic diagnosis platform, breath analysis is gaining much attention because it is a noninvasive, simple, and low cost diagnostic method. Very promising clinical applications have been demonstrated for diagnostic purposes by correlation analysis between exhaled breath components and specific diseases. In addition, diverse breath molecules, which serve as biomarkers for specific diseases, are precisely identified by statistical pattern recognition studies. To further improve the accuracy of breath analysis as a diagnostic tool, breath sampling, biomarker sensing, and data analysis should be optimized. In particular, development of high performance breath sensors, which can detect biomarkers at the ppb-level in exhaled breath, is one of the most critical challenges. Due to the presence of numerous interfering gas species in exhaled breath, selective detection of specific biomarkers is also important. This Account focuses on chemiresistive type breath sensors with exceptionally high sensitivity and selectivity that were developed by combining hollow protein templated nanocatalysts with electrospun metal oxide nanostructures. Nanostructures with high surface areas are advantageous in achieving high sensitivity because the sensing signal is dominated by the surface reaction between the sensing layers and the target biomarkers. Furthermore, macroscale pores between one-dimensional (1D) nanostructures can facilitate fast gas diffusion into the sensing layers. To further enhance the selectivity, catalytic functionalization of the 1D metal oxide nanostructure is essential. However, the majority of conventional techniques for catalytic functionalization have failed to achieve a high degree of dispersion of nanoscale catalysts due to aggregation on the surface of the metal oxide, which severely deteriorates the sensing properties by lowering catalytic activity. This issue has led to extensive studies on monolithically dispersed nanoscale particles on metal oxides to maximize the catalytic performances. As a pioneering technique, a bioinspired templating route using apoferritin, that is, a hollow protein cage, has been proposed to obtain nanoscale (∼2 nm) catalyst particles with high dispersity. Nanocatalysts encapsulated by a protein shell were first used in chemiresistive type breath sensors for catalyst functionalization on 1D metal oxide structures. We discuss the robustness and versatility of the apoferrtin templating route for creating highly dispersive catalytic NPs including single components (Au, Pt, Pd, Rh, Ag, Ru, Cu, and La) and bimetallic catalysts (PtY and PtCo), as well as the core-shell structure of Au-Pd (Au-core@Pd-shell). The use of these catalysts is essential to establish high performance sensors arrays for the pattern recognition of biomarkers. In addition, novel multicomponent catalysts provide unprecedented sensitivity and selectivity. With this in mind, we discuss diverse synthetic routes for nanocatalysts using apoferritin and the formation of various catalyst-1D metal oxide composite nanostructures. Furthermore, we discuss detection capability of a simulated biomarker gas using the breath sensor arrays and principal component analysis. Finally, future prospects with the portable breath analysis platform are presented by demonstrating the potential feasibility of real-time and on-site breath analysis using chemiresistive sensors.

Knowledge graph streams are a data model underlying many online dynamic data applications today. Answering predictive relationship queries over such a stream is very challenging as the heterogeneous graph streams imply complex topological and temporal correlations of knowledge facts, as well as fast dynamic incoming rates and statistical pattern changes over time. We present our approach with two major components: a Count-Fading sketch and an online incremental embedding algorithm. We answer predictive relationship queries using the embedding results. Extensive experiments over real world datasets show that our approach significantly outperforms two baseline approaches, producing accurate query results efficiently with a small memory footprint.

Data-driven damage localization is an important step of vibration-based structural health monitoring. Statistical pattern recognition based on the prominent steps of feature extraction and statistical decision-making provides an effective and efficient framework for structural health monitoring. However, these steps may become time-consuming or complex when there are large volumes of vibration measurements acquired by dense sensor networks. To deal with this issue, this study proposes fast unsupervised learning methods for feature extraction through autoregressive modeling and damage localization through a new distance measure called Kullback–Leibler divergence with empirical probability measure. The feature extraction approach consists of an iterative algorithm for order selection and parameter estimation aiming to extract residuals in the training phase and another iterative process aiming to extract residuals only in the monitoring phase. The key feature of the proposed approach is the use of correlated residual samples of the autoregressive model as a new time series at each iteration, rather than handling the measured vibration response of the structure. This is shown to highly reduce the computational burden of order selection and feature extraction; moreover, it effectively provides low-order autoregressive models with uncorrelated residuals. The Kullback–Leibler divergence with empirical probability measure method exploits a segmentation technique to subdivide random data into independent sets and provides a distance metric based on the theory of empirical probability measure with no need to explicitly compute the actual probability distributions at the training and monitoring stages. Numerical and experimental benchmarks are then used to assess accuracy and performance of the proposed methods and compare them with some state-of-the-art approaches. Results show that the proposed approaches are successful in feature extraction and damage localization, with a reduced computational burden.

Surface structuring by laser processing is a key technology enabler for a wide range of surface functionalization applications. Statistical recognition of linear geometrical features produced on initially rough surfaces is developed and applied demonstrating reliable analysis of microstructures., Lasers Applications Conference 2020, Oct 13-16, 2020 , Washington, D.C.

Arbitrage is one of the most popular strategies in the spectrum of low-risk stock market tactics. Professional arbitrage traders are market participants that form more than 80 percent of market patterns. For a novice trader, these patterns are invisible. But some trading models indirectly rely on what arbitrage creates. The article provides an overview of the basic scenarios of arbitration in the Russian stock market and describes its own technology that is not tied to the terms of the contract with the exchange. This strategy is based on a statistical pattern. The article gives an overview of the basic arbitration scenarios on the Russian stock market and describes the own technology not tied to the terms of the contract with the stock exchange. This strategy is based on statistical regularity and allows to gain an advantage over some categories of market participants. Software has been developed to search for price differences between the same assets that are traded on different exchanges, or to find different assets that have a statistical relationship. The main advantage of the developed system is the speed, since the average duration of the profitable situation is less than 5 seconds.

In order to improve the operational safety of NPPs and improve the support systems for operating personnelof nuclear power units and exclusion of possible errors due to human factors, as part of the developed automatic system for operational diagnostics of the current technical condition of the main equipment (in particular, themain aggregate nodes of the main primary circulation pumps), an additional subsystem for a short-term forecast of the trend of changes in this state was created. The forecast subsystem is designed to enable the earlydetection of latent initial phases of potentially dangerous violations of the normal course of physical processesin the most vulnerable nodes of pumping units and the timely prevention of the development of identified deviations and anomalies in emergency situations. The functional core of the diagnostics and prognosis automaticsystem was developed on the basis of the theory of statistical pattern recognition. The forecast subsystem isbased on a model and an algorithm that extrapolate the current series of previous values for the nearest futuretime period to the generalized statistical criterion proposed by the authors, which is used by the developed automatic operational diagnostics system to assess the current technical condition of pumping units and representsthe total a posteriori probability of the technical states of the monitored units. The use of the system will allowplanning repair operations in accordance with the actual technical condition of the equipment and therebyreduce the economic losses from unnecessary downtime.

Over the past two decades, statistical methods have played a prominent role in the development of pattern recognition techniques. The statistical decision theory and related fields have been a forum where significant theoretical advances and innovations have taken place. This chapter presents a brief and elementary account of some important topics in the statistical theory of pattern recognition. Statistical methods provide the proper framework for studying pattern recognition problems when the pattern-generating mechanism can be faithfully represented by a statistical model, and the goal of the recognition is to decide whether or not a given pattern belongs to some prespecified class of patterns. People merely have some vague idea about the underlying statistical distributions, together with a number of design samples. It is therefore necessary to give the pattern-recognition system, the capability of learning the unknown information from the data available.

Traditional profiling techniques typically identify performance hot-spots. Other specialized techniques such as WCET analysis cater to safety critical real-time systems with hard constraints and thus make conservative assumptions. However, several domains motivate the need to systematically characterize and represent differential timing properties such as the variance in execution time of application artifacts such as the functions. Such domains include, vulnerability analysis of applications with regard to differential timing attacks, and the optimization of soft-real-time applications to reduce frame-rate fluctuations. In this paper, we motivate the need for execution variance as a performance measure and propose a variance-based analysis scheme. We introduce a new program representation called Variance Characterization Graph (VCG) that is used both as the intermediate representation for the variance-based analysis, and as the final representation that provides concise actionable information to programmers and optimization frameworks. We develop a methodology based on statistical pattern matching to summarize the dominant patterns of application behavior into a very compact VCG representation useful for tuning application behavior such as the soft-real-time properties.

Selection of optimal treatment strategy for a cracked asphalt pavement requires efficient evaluation of cracking. Pavement cracking severity was usually evaluated just by crack width in the past. I...

Many research works based on the tone of human skin have been developed to locate and track the human body for the purpose of recognition in color images. With respect to other techniques, some advantages of face detection based on skin color are the smaller processing time, invariant angles of rotation and the performance in semi-occluded faces. In this study we present the results of a survey that investigated the performance of 4 supervised classifiers in skin detection. In order to maximize the generalization of the models, a training set containing samples of individuals of different ages and ethnicities was used. Experimental results showed that the best performance was achieved by using an ANN and the worst results were yielded by LDA. With the Naive Bayes, QDA and ANN algorithms, we showed that the white, black, yellow and brown tones of human skin are in a well-defined range of the RGB color spectrum determined by common characteristics. We also compiled 2798 skin samples for treatment and 305 images with their manually obtained labels as supplementary material, which was made available to help in the development of further research in human skin detection.

The current review investigates the question of the relationship of different hemispheric specializations. Hemispheric specializations are the result of the seemingly distinctive ability of the left and right hemisphere to specialize in different cognitive functions. The review focuses on the concept of complementarity whereby the lateralization of one function predicts the asymmetric processing of another one. The complementarity of hemispheric specializations presents an interesting dilemma as the literature evidence is outwardly mixed. The causal and statistical hypotheses attempt to account for the observed findings, with support for both hypotheses. However, most of the evidence appears to align with the statistical pattern of complementarity. Converging lines of evidence suggest that there are multiple independent biases that play a role in how these functional cerebral asymmetries are organized in the human brain. Some of the functional and evolutionary implications of the existence of complementarity of hemispheric specializations are also briefly discussed.

A method for localization and severity assessment of structural damages in concrete dams is proposed. The algorithm works are based on observed displacement data that respond from the structures. Statistical pattern recognition based structural damage detection is often developed exploiting the methods of inversion analysis. The relationship between the observed displacement and damage identification of concrete dams is studied. By using the inversion method, the functional relationship between the observed displacement and damage information is developed to calculate the damage defining parameters, which are damage location and severity, are updated through an optimization scheme. Meanwhile, the damage index of a concrete dam is determined with these methods and principles. These indexes have been applied into the safety monitoring of that dam and satisfactory result has been achieved.

In recent years the number of new malware threats has increased significantly, causing a damage of billions of dollars globally. To counter this aggressive malware attack, the anti-malware industry needs to be able to correctly classify malware in order to provide defense against them. Consequently, malware classification has been an active area of research, and a multitude of malware classification approaches have been proposed in the literature. This paper evaluates two methods of sequence classification based on Hidden Markov Model, namely the maximum likelihood and similarity-based methods, for classification of malware using a large and comprehensive dataset. System calls generated by known malware during execution are used as observation sequences to train the Hidden Markov Models. Malware samples are evaluated against the trained models to produce similarity vectors, which are used in the maximum likelihood and similarity-based classification schemes to predict the family for an unknown malware sample. Comparison of the two schemes shows that combining the powerful statistical pattern analysis capability of Hidden Markov Models and discriminative classifiers in the similarity- based method results in a significantly better classification performance as compared to the maximum likelihood approach. Furthermore, evaluation of different classifiers in the similarity-based method demonstrates that Random Forest classifier performs better than other classifiers on malware similarity vectors.

Structural conditions assessment utilizes data analysis methods based on signals measured from subject structures. Among the various signal-based analysis methods available, a Statistical Pattern-recognition Technique (SPRT) based on Mahalanobis Distance (MD) theory is considered useful in assessing structural conditions; however, its evaluation of volatile signal data incurred by external loads such as seismic loads is usually limited. In this study, an experiment was conducted to attempt to overcome such limitations of MD theory. Furthermore, research was conducted on condition assessment methods based on SPRTs that can accurately assess structural conditions, even with highly volatile measured data. As a first step, the limitations of the standard MD theory were experimentally confirmed. Based on these limitations, a revised MD (RMD) theory was subsequently proposed. To verify the proposed SPRT, standardized cyclic loads and non-standardized seismic loads were used to conduct experiments on structural damage assessment. Based on these results, the RMD theory proposed in this study was confirmed to be a valid method for comprehending structural damage.

In this proof of concept study, chemical threat agent (CTA) samples were classified to their sources with accuracies of 87-100% by applying a random forest statistical pattern recognition technique to analytical data acquired by comprehensive two-dimensional gas chromatography with time-of-flight mass spectrometric detection (GC × GC-TOFMS). Three organophosphate pesticides, chlorpyrifos, dichlorvos, and dicrotophos, were used as the model CTAs, with data collected for 4-6 sources per CTA and 7-10 replicate analyses per source. The analytical data were also evaluated to determine tentatively identified chemical attribution signatures for the CTAs by comparing samples from different sources according to either the presence/absence of peaks or the relative responses of peaks. These results demonstrate that GC × GC-TOFMS analysis in combination with a random forest technique can be useful in sample classification and signature identification for pesticides. Furthermore, the results suggest that this combination of analytical chemistry and statistical approaches can be applied to forensic analysis of other chemicals for similar purposes.