Statistical Classification Scheme Research Articles

Social Division of Labour as an Explanatory Principle and as a Foundation for Legitimation

The article presents results of the analysis of the conceptions and uses of the social division of labour by the statistical classification schemes and by the social sciences. The general hypothesis pursued is that the social division of labour is used as an explanatory principle and as a basis for legitimation of the social structures. These uses encompass both the current statistical classification schemes and the social sciences involved. In addition to the introduction, a second item discusses the origins and meanings of the social division of labour for the social sciences and its relations with theoretical problems such as the multidimensionality of social structures. Next, the synthesis of the conception and uses of the social division of labour by the main statistical classification schemes in vogue is presented. The high degree of redundancy of statistical classifications is exposed below, particularly with regard to occupations and status in employment. The categories of managers are particularly examined as the main empirical reference. The next item presents the results of the examination of the overlapping of categories linked management with the condition of the employer and the overvaluation of the amount of schooling. Finally, the last item discusses general issues related to the uses of the social division of labour and its limits and relationships with theoretical foundations of the social sciences.

Statistical network protocol identification with unknown pattern extraction

Network traffic classification is an enabling technique for network security and management for both traditional networks and emerging networks such as Internet of Things. Due to the decreasing effectiveness of traditional port-based and payload-based methods, lots of research attentions are devoted to an alternative approach based on flow and packet-level traffic characteristics. A variety of statistical classification schemes are proposed in this context, but most of them embody an implicit assumption that all protocols are known in advance and well presented in the training data. This assumption is unrealistic because real-world networks constantly witness emerging traffic patterns and protocols that are previously unknown. In this paper, we revisit the problem by proposing a learning scheme with unknown pattern extraction for statistical protocol identification. The scheme is designed with a more realistic setting, in which we assume that the training data only consists of labeled samples from a limited number of protocols, and the goal is to identify these known patterns out of arbitrary traffic mixture of both known and unknown protocols. Our experiments based on real-world traffic show that the proposed scheme outperforms previous approaches by accurately identifying both known and unknown protocols.

Model-free fault detection and isolation of a benchmark process control system based on multiple classifiers techniques—A comparative study

This paper presents a combined data-driven framework for fault detection and isolation (FDI) based on the ensemble of diverse classification schemes. The proposed FDI scheme is configured in series and parallel forms in the sense that in series form the decision on the occurrence of fault is made in FD module, and subsequently, the FI module coupled to the FD module will be activated for fault indication purposes. On the other hand, in parallel form a single module is employed for FDI purposes, simultaneously. In other words, two separate multiple-classifiers schemes are presented by using fourteen various statistical and non-statistical classification schemes. Furthermore, in this study, a novel ensemble classification scheme namely blended learning (BL) is proposed for the first time where single and boosted classifiers are blended as the local classifiers in order to enrich the classification performance. Single-classifier schemes are also exploited in FDI modules along with the ensemble-classifier methods for comparison purposes. In order to show the performance of proposed FDI method, it was also tested and validated on DAMADICS actuator system benchmark. Besides, comparative study with the related works done on this benchmark is provided to show the pros and cons of the proposed FDI method.

Repeat or not repeat?—Statistical validation of tandem repeat prediction in genomic sequences

Tandem repeats (TRs) represent one of the most prevalent features of genomic sequences. Due to their abundance and functional significance, a plethora of detection tools has been devised over the last two decades. Despite the longstanding interest, TR detection is still not resolved. Our large-scale tests reveal that current detectors produce different, often nonoverlapping inferences, reflecting characteristics of the underlying algorithms rather than the true distribution of TRs in genomic data. Our simulations show that the power of detecting TRs depends on the degree of their divergence, and repeat characteristics such as the length of the minimal repeat unit and their number in tandem. To reconcile the diverse predictions of current algorithms, we propose and evaluate several statistical criteria for measuring the quality of predicted repeat units. In particular, we propose a model-based phylogenetic classifier, entailing a maximum-likelihood estimation of the repeat divergence. Applied in conjunction with the state of the art detectors, our statistical classification scheme for inferred repeats allows to filter out false-positive predictions. Since different algorithms appear to specialize at predicting TRs with certain properties, we advise applying multiple detectors with subsequent filtering to obtain the most complete set of genuine repeats.

Applying simple water-energy balance frameworks to predict the climate sensitivity of streamflow over the continental United States

Abstract. The prediction of climate effects on terrestrial ecosystems and water resources is one of the major research questions in hydrology. Conceptual water-energy balance models can be used to gain a first order estimate of how long-term average streamflow is changing with a change in water and energy supply. A common framework for investigation of this question is based on the Budyko hypothesis, which links hydrological response to aridity. Recently, Renner et al. (2012) introduced the climate change impact hypothesis (CCUW), which is based on the assumption that the total efficiency of the catchment ecosystem to use the available water and energy for actual evapotranspiration remains constant even under climate changes. Here, we confront the climate sensitivity approaches (the Budyko approach of Roderick and Farquhar, 2011, and the CCUW) with data of more than 400 basins distributed over the continental United States. We first estimate the sensitivity of streamflow to changes in precipitation using long-term average data of the period 1949 to 2003. This provides a hydro-climatic status of the respective basins as well as their expected proportional effect to changes in climate. Next, we test the ability of both approaches to predict climate impacts on streamflow by splitting the data into two periods. We (i) analyse the long-term average changes in hydro-climatology and (ii) derive a statistical classification of potential climate and basin change impacts based on the significance of observed changes in runoff, precipitation and potential evapotranspiration. Then we (iii) use the different climate sensitivity methods to predict the change in streamflow given the observed changes in water and energy supply and (iv) evaluate the predictions by (v) using the statistical classification scheme and (vi) a conceptual approach to separate the impacts of changes in climate from basin characteristics change on streamflow. This allows us to evaluate the observed changes in streamflow as well as to assess the impact of basin changes on the validity of climate sensitivity approaches. The apparent increase of streamflow of the majority of basins in the US is dominated by an increase in precipitation. It is further evident that impacts of changes in basin characteristics appear simultaneously with climate changes. There are coherent spatial patterns with catchments where basin changes compensate for climatic changes being dominant in the western and central parts of the US. A hot spot of basin changes leading to excessive runoff is found within the US Midwest. The impact of basin changes on the prediction is large and can be twice as much as the observed change signal. Although the CCUW and the Budyko approach yield similar predictions for most basins, the data of water-limited basins support the Budyko framework rather than the CCUW approach, which is known to be invalid under limiting climatic conditions.

Multi-scaled morphological features for the characterization of mammographic masses using statistical classification schemes

A comprehensive signal analysis approach on the mammographic mass boundary morphology is presented in this article. The purpose of this study is to identify efficient sets of simple yet effective shape features, employed in the original and multi-scaled spectral representations of the boundary, for the characterization of the mammographic mass. These new methods of mass boundary representation and processing in more than one domain greatly improve the information content of the base data that is used for pattern classification purposes, introducing comprehensive spectral and multi-scale wavelet versions of the original boundary signals. The evaluation is conducted against morphological and diagnostic characterization of the mass, using statistical methods, fractal dimension analysis and a wide range of classifier architectures. This study consists of (a) the investigation of the original radial distance measurements under the complete spectrum of signal analysis, (b) the application of curve feature extractors of morphological characteristics and the evaluation of the discriminative power of each one of them, by means of statistical significance analysis and dataset fractal dimension, and (c) the application of a wide range of classifier architectures on these morphological datasets, in order to conduct a comparative evaluation of the efficiency and effectiveness of all architectures, for mammographic mass characterization. Radial distance signal was exploited using the discrete Fourier transform (DFT) and the discrete wavelet transform (DWT) as additional carrier signals. Seven uniresolution feature functions were applied over these carrier signals and multiple shape descriptors were created. Classification was conducted against mass shape type and clinical diagnosis, using a wide range of linear and non-linear classifiers, including linear discriminant analysis (LDA), least-squares minimum distance (LSMD), k-nearest neighbor (k-NN), radial basis function (RBF) and multi-layered perceptron (MLP) neural networks (NN), and support vector machines (SVM). Fractal analysis was employed as a dataset analysis tool in the feature selection phase. The discriminative power of the features produced by this composite analysis is subsequently analyzed by means of multivariate analysis of variance (MANOVA) and tested against two distinct classification targets, namely (a) the morphological shape type of the mass and (b) the histologically verified clinical diagnosis for each mammogram. Statistical analysis and classification results have shown that the discrimination value of the features extracted from the DWT components and especially the DFT spectrum, are of great importance. Furthermore, much of the information content of the curve features in the case of DFT and DWT datasets is directly related to the texture and fine-scale details of the corresponding envelope signal of the spectral components. Neural classifiers outperformed all other methods (SVM not used because they are mainly two-class classifiers) with overall success rate of 72.3% for shape type identification, while SVM achieved the overall highest 91.54% for clinical diagnosis. Receiver operating characteristic (ROC) analysis has been employed to present the sensitivity and specificity of the results of this study.

Race or People: Federal Race Classifications for Europeans in America, 1898--1913

In 1898 the U.S. Bureau of Immigration initiated a classification of immigrants into some 40 categories of or people; nearly all the categories covered Europeans. In 1909 an effort was made to extend this system of classification to the U.S. Census, and the relevant measure passed in the Senate. From the outset, organizations representing a segment of American Jews strongly opposed the measure, although not on the grounds of racism. But other groups of immigrants, including Jews, strongly supported the new racial classification of Europeans for the census. A compromise replaced the proposed new race question with a mother-tongue question. The paper explains the origin and development of the classification system and the ensuing controversy; extensive verbatim transcripts (in which participants argue their conception of race in the context of other terms) and unpublished letters constitute the basic sources. The or people classification was immensely important in its own right, since our knowledge of the socioeconomic characteristics of immigrants in the first half of the 20th century is organized in terms of that classification. But the topic is interesting for much broader reasons: discussion of a seemingly narrow and technical matter, namely a statistical classification scheme, illuminates the meaning of race for the debaters and sheds light on the dynamics of ideas, bureaucracy, and organized opposition to official procedures.

High-Emitting Vehicle Characterization Using Regression Tree Analysis

A small fraction of motor vehicles on the roadway emit a disproportionate fraction of pollutant emissions, especially for carbon monoxide and hydrocarbons. Generally, these “high emitters” or “super emitters” exhibit higher emissions rates under all operating conditions than do “normal emitters.” Since the instantaneous emissions response between normal- and high-emitting vehicles can differ by one or more orders of magnitude, so do their average emissions over a “typical” trip. Identifying the proportion of normal- and high-emitting vehicles in an urban area and quantifying their emissions is vital for accurate emission inventory accounting. A methodology by which high and normal emitters can be classified is presented. Unlike previous emitter classification approaches, the approach is data driven and relies entirely on hot-stabilized emissions results. A statistical classification scheme, better known as hierarchical tree based regression, is used to separate vehicles into homogenous emitter categories. The approach is shown to have a number of advantages. First, it is flexible with respect to both the number of classes and types of variables used to identify classes. Second, it considers the influence of a large number of vehicle and technology attributes on emitter status. Third, it ensures that the highest emitters can be isolated from the normal emitters, so that separate emission rate models can be developed for these vehicles. Finally, the approach does not combine the effects of starts and hot-stabilized operations within the definition of high emitter, leading to a classification scheme whereby vehicles with poor start emissions characteristics will not be incorrectly classified as vehicles with poor hot-stabilized emission characteristics.

CLASSIFICATION THEORY FOR PHASE TRANSITIONS

A refined classification theory for phase transitions in thermodynamics and statistical mechanics in terms of their orders is introduced and analyzed. The refined thermodynamic classification is based on two independent generalizations of Ehrenfests traditional classification scheme. The statistical mechanical classification theory is based on generalized limit theorems for sums of random variables from probability theory and the newly defined block ensemble limit. The block ensemble limit combines thermodynamic and scaling limits and is similar to the finite size scaling limit. The statistical classification scheme allows for the first time a derivation of finite size scaling without renormalization group methods. The classification distinguishes two fundamentally different types of phase transitions. Phase transitions of order λ>1 correspond to well known equilibrium phase transitions, while phase transitions with order λ<1 represent a new class of transitions termed anequilibrium transitions. The generalized order λ varies inversely with the strength of fluctuations. First order and second order transitions play a special role in both classification schemes. First order transitions represent a limiting case separating equilibrium and anequilibrium transitions. The special role or second order transitions is shown to be related to the breakdown of hyperscaling. For anequilibrium transitions the nature of the heat bath in the canonical ensemble becomes important.

Classification theory for anequilibrium phase transitions

The paper introduces a classification of phase transitions in which each transition is characterized through its generalized order and a slowly varying function. This characterization is shown to be applicable in statistical mechanics as well as in thermodynamics albeit for different mathematical reasons. By introducing the block ensemble limit the statistical classification is based on the theory of stable laws from probability theory. The block ensemble limit combines scaling limit and thermodynamic limit. The thermodynamic classification on the other hand is based on generalizing Ehrenfest's traditional classification scheme. Both schemes imply the validity of scaling at phase transitions without the need to invoke renormalizaton-group arguments. The statistical classification scheme allows derivation of a form of finite-size scaling for the distributions of statistical averages while the thermodynamic classification gives rise to multiscaling of thermodynamic potentials. The different nature of the two classification theories is also apparent from the fact that the generalized thermodynamic order is unbounded while the statistical order is restricted to values less than 2. This fact is found to be related to the breakdown of hyperscaling relations. Both classification theories predict the possible existence of phase transitions having orders less than unity. Such transitions are termed anequilibrium transitions.Systems near anequilibrium transitions cannot be described by conventional equilibrium thermodynamics or equilibrium statistical mechanics because of very strong fluctuations. Anequilibrium transitions are found to exist in statistical-mechanical model systems. The identification of the Lagrange parameter \ensuremath{\beta} in the canonical ensemble becomes invalid if a reservoir and a system of the same substance are in thermal contact and anequilibrium transitions are present. Based on the ergodic hypothesis and the theory of convolution semigroups it is shown that near anequilibrium transitions the equations of motion for macroscopic observables of infinite systems may involve modified time derivatives as generators of the macroscopic time evolution. The general solution to the modified equations of motion exhibits very slow dynamics as frequently observed in a nonequilibrium experiment.

Predicting chemically induced duodenal ulcer and adrenal necrosis with classification trees.

Binary tree-structured statistical classification algorithms and properties of 56 model alkyl nucleophiles were brought to bear on two problems of experimental pharmacology and toxicology. Each rat of a learning sample of 745 was administered one compound and autopsied to determine the presence of duodenal ulcer or adrenal hemorrhagic necrosis. The cited statistical classification schemes were then applied to these outcomes and 67 features of the compounds to ascertain those characteristics that are associated with biologic activity. For predicting duodenal ulceration, dipole moment, melting point, and solubility in octanol are particularly important, while for predicting adrenal necrosis, important features include the number of sulfhydryl groups and double bonds. These methods may constitute inexpensive but powerful ways to screen untested compounds for possible organ-specific toxicity. Mechanisms for the etiology and pathogenesis of the duodenal and adrenal lesions are suggested, as are additional avenues for drug design.

Recognition of handwritten symbols

New statistical methods are presented for recognizing isolated, handwritten or printed symbols directly from raster images of documents such as technical drawings and maps. Our methods for recognizing symbols of known or unknown rotation avoid the traditional thinning and vectorization steps in the recognition process. Hence, the information loss caused by these two steps is eliminated. In all of our approaches we consider the outer pixel boundary of an isolated symbol candidate in the binary raster image as a simple closed curve. This curve is then approximated by a parametric spline curve, an elliptic Fourier expansion of the Fourier expansion due to Zahn and Roskies. Curvature values and coordinates along the spline curves or the coefficients of the Fourier expansion are then used as descriptors in a statistical classification scheme. Statistical classification using these methods was done on separate training and test sets consisting of the digits 0 to 9 and all lower-case letters. There were about one hundred samples of each handwritten symbol class in each set. The correct classification rate was as high as 97.7% using the parametric spline approximation, and 98.6% using the elliptic Fourier descriptors.

A nonparametric classification scheme with mean squared error criterion

A statistical classification scheme for a given set of data requires knowledge of the probability distribution of the observations. Traditional approaches to this problem have revolved around chosen various parametric forms for the probability distribution and evaluating these by goodness of fit methods. Among the difficulties with this method are that it is time consuming, it may not lead to satisfactory results and it may lie beyond the statistical expertise of many practitioners. In this paper, the author's consider the use of a recently developed nonparametric probability density estimator in classification schemes with mean squared error loss criterion. Classical parametric approaches are compared to the nonparametric method on simulated data on the basis of the misclassification probability. Real data from the medical and biological sciences are also used to illustrate the usefulness of the nonparametric method.