Average Correct Classification Rate Research Articles

Efficient Color Texture Classification Using Color Monogenic Wavelet Transform

Color textures are among the most important visual attributes in image analysis. From the practical point view of color texture image analysis, this paper proposes an effective multi-scale color texture classification algorithm that is rotation and scale invariant using non-marginal color monogenic wavelet transform. The proposed algorithm exploits the color monogenic wavelet transform to obtain multi-scale representation of training samples for each texture class. The coefficients of color monogenic wavelet transform represent a magnitude and three phases: two phases encode local color information while the third contains geometric information of color texture image. The multi-scale feature vector is composed of mean value, standard deviation, energy and entropy at different scales of each of the directional sub-bands. The experimental results of average correct classification rates are 98.67, 99.08 and $$99.89\%$$ which are obtained from different color texture databases demonstrate its superior performance and robustness of the proposed classifier. The proposed color texture feature vector is also shown to be effective for color texture classification.

Selective Fusion of Heterogeneous Classifiers for Predicting Substrates of Membrane Transporters.

Membrane transporters play a crucial role in determining fate of administered drugs in a biological system. Early identification of plausible transporters for a drug molecule can provide insights into its therapeutic, pharmacokinetic, and toxicological profiles. In the present study, predictive models for classifying small molecules into substrates and nonsubstrates of various pharmaceutically important membrane transporters were developed using quantitative structure-activity relationship (QSAR) and proteochemometric (PCM) approaches. For this purpose, 4575 substrate interactions for these transporters were collected from the Metabolism and Transport Database (Metrabase) and the literature. The transporters selected for this study include (i) six efflux transporters, viz., breast cancer resistance protein (BCRP/ABCG2), P-glycoprotein (P-gp/MDR1), and multidrug resistance proteins (MRP1, MRP2, MRP3, and MRP4), and (ii) seven influx transporters, viz., organic cation transporter (OCT1/SO22A1), peptide transporter (PEPT1/SO15A1), apical sodium-bile acid transporter (ASBT/NTCP2), and organic anion transporting peptides (OATP1A2/SO1A2, OATP1B/SO1B1, OATP1B3/SO1B3, and OATP2B1/SO2B1). Various types of descriptors and machine learning methods (classifiers) were evaluated for the development of robust predictive models. Additionally, ensemble models were developed by bagging of homogeneous classifiers and selective fusion of heterogeneous classifiers. It was observed that the latter approach improves the accuracy of substrate/nonsubstrate prediction for transporters (average correct classification rate of more than 0.80 for external validation). Moreover, structural fragments important in determining the substrate specificity across the various transporters were identified. To demonstrate these fragments on the query molecule, contour maps were generated. The prediction efficacy of the developed models was illustrated by a good correlation between the reported logBB value of a molecule and its predicted substrate propensity for blood-brain barrier transporters. Conclusively, this comprehensive modeling analysis can be efficiently employed for the prediction of membrane transporters of a drug, thereby providing insights into its pharmacological profile.

Loan decision models for the Jordanian commercial banks

Credit risk remains the most critical challenge facing bank's management as it adversely affects the profitability and stability of the bank. However, despite the rise in loan delinquency and the serious competition in the banking market, loan application evaluations at the Jordanian commercial banks are subjective in nature. Additionally, the ability to discriminate between 'good' and 'bad' risk applications is critical. Rejecting a good application might cause loss of future potential profit while approving a bad application might cause loss of principal money and interest. The current research aims to develop credit decision support using linear discriminant analysis (LDA), multi-layer perceptron (MLP) and CART decision trees for the protection against credit risk. A pooled data set of personal loans from Jordanian commercial banks was used to build the decision models. The discriminative power of the developed models was assessed using average correct classification rate (ACC) and the estimated misclassification cost (EMC). The results showed that the MLP model achieved the highest ACC as well as the lowest EMC.

Classification of Bacillus and Brevibacillus species using rapid analysis of lipids by mass spectrometry.

Bacillus are aerobic spore-forming bacteria that are known to lead to specific diseases, such as anthrax and food poisoning. This study focuses on the characterization of these bacteria by the detection of lipids extracted from 33 well-characterized strains from the Bacillus and Brevibacillus genera, with the aim to discriminate between the different species. For the purpose of analysing the lipids extracted from these bacterial samples, two rapid physicochemical techniques were used: matrix-assisted laser desorption/ionisation time-of-flight mass spectrometry (MALDI-TOF-MS) and liquid chromatography in conjunction with mass spectrometry (LC-MS). The findings of this investigation confirmed that MALDI-TOF-MS could be used to identify different bacterial lipids and, in combination with appropriate chemometrics, allowed for the discrimination between these different bacterial species, which was supported by LC-MS. The average correct classification rates for the seven species of bacteria were 62.23 and 77.03 % based on MALDI-TOF-MS and LC-MS data, respectively. The Procrustes distance for the two datasets was 0.0699, indicating that the results from the two techniques were very similar. In addition, we also compared these bacterial lipid MALDI-TOF-MS profiles to protein profiles also collected by MALDI-TOF-MS on the same bacteria (Procrustes distance, 0.1006). The level of discrimination between lipids and proteins was equivalent, and this further indicated the potential of MALDI-TOF-MS analysis as a rapid, robust and reliable method for the classification of bacteria based on different bacterial chemical components. Graphical abstractMALDI-MS has been successfully developed for the characterization of bacteria at the subspecies level using lipids and benchmarked against HPLC Electronic supplementary materialThe online version of this article (doi:10.1007/s00216-016-9890-4) contains supplementary material, which is available to authorized users.

Identification of pesticide varieties by testing microalgae using Visible/Near Infrared Hyperspectral Imaging technology.

In our study, the feasibility of using visible/near infrared hyperspectral imaging technology to detect the changes of the internal components of Chlorella pyrenoidosa so as to determine the varieties of pesticides (such as butachlor, atrazine and glyphosate) at three concentrations (0.6 mg/L, 3 mg/L, 15 mg/L) was investigated. Three models (partial least squares discriminant analysis combined with full wavelengths, FW-PLSDA; partial least squares discriminant analysis combined with competitive adaptive reweighted sampling algorithm, CARS-PLSDA; linear discrimination analysis combined with regression coefficients, RC-LDA) were built by the hyperspectral data of Chlorella pyrenoidosa to find which model can produce the most optimal result. The RC-LDA model, which achieved an average correct classification rate of 97.0% was more superior than FW-PLSDA (72.2%) and CARS-PLSDA (84.0%), and it proved that visible/near infrared hyperspectral imaging could be a rapid and reliable technique to identify pesticide varieties. It also proved that microalgae can be a very promising medium to indicate characteristics of pesticides.

Classification based hypothesis testing in neuroscience: Below‐chance level classification rates and overlooked statistical properties of linear parametric classifiers

Multivariate pattern analysis (MVPA) has recently become a popular tool for data analysis. Often, classification accuracy as quantified by correct classification rate (CCR) is used to illustrate the size of the effect under investigation. However, we show that in low sample size (LSS), low effect size (LES) data, which is typical in neuroscience, the distribution of CCRs from cross-validation of linear MVPA is asymmetric and can show classification rates considerably below what would be expected from chance classification. Conversely, the mode of the distribution in these cases is above expected chance levels, leading to a spuriously high number of above chance CCRs. This unexpected distribution has strong implications when using MVPA for hypothesis testing. Our analyses warrant the conclusion that CCRs do not well reflect the size of the effect under investigation. Moreover, the skewness of the null-distribution precludes the use of many standard parametric tests to assess significance of CCRs. We propose that MVPA results should be reported in terms of P values, which are estimated using randomization tests. Also, our results show that cross-validation procedures using a low number of folds, e.g. twofold, are generally more sensitive, even though the average CCRs are often considerably lower than those obtained using a higher number of folds. Hum Brain Mapp 37:1842-1855, 2016. © 2016 Wiley Periodicals, Inc.

Identifying Irregular Potatoes by Developing an Intelligent Algorithm Based on Image Processing

The objective of this study was to develop an algorithm based on image processing for detecting misshapen potatoes from the mass of potatoes and obtaining homogeneous products. The database used in this research included the digital images acquired from Agria variety of Ardabil (Iranian northern-west) potato with different sizes and shapes. A combination of morphological features including geometrical features like length, width and features related to shape such as roundness were taken into consideration in identifying irregular potatoes from others employing elongation and Fourier descriptors. Using statistical principal component analysis (PCA), seven features were selected as the most prominent for classification. The experimental results showed that the proposed method achieves a high level of accuracy with merely seven selected discriminative features, obtaining an average correct classification rate of 98% for training set. Additionally, regular potatoes were separated into small, medium and large categories with 100% accuracy. According to the results, the developed algorithm based on image processing can be used in classifying products with no proper shape.

Deep neural network for halftone image classification based on sparse auto-encoder

To restore high quality continuous tone images from each class of halftone images, halftone image fine classification is the key problem. In this paper, a novel feature learning method is proposed for classifying 14 kinds of halftone images produced by the most well-known halftoning algorithms. This study employs the stacked sparse auto-encoders (SAE) trained with unsupervised learning for extracting features of halftone images, and then uses softmax regression with supervised learning for fine-tuning the deep neural network and classifying halftone images. In order to reduce the run-time of deep neural network and improve the image correct classification rate, we propose an effective patch extraction method for testing halftone images by measuring the mean and variance of local entropy in a patch. Halftone image classification is determined by the classification results of all effective patches inside an image via majority voting (MV). The experimental results demonstrate that our proposed method achieves an average correct classification rate (ACCR) of over 99.44% for 14 kinds of halftone images on two public image sets. Compared with state-of-the-art LMS–Bayes and M10–ML methods, the proposed SAE-MV method can distinguish the most categories of halftone images and achieve competitive ACCR, meanwhile, demonstrate better generalization performance.

Identifying irregular potatoes by developing an intelligent algorithm based on image processing

The objective of this study was to develop an algorithm based on image processing for detecting misshapen potatoes from the mass of potatoes and obtaining homogeneous products. The database used in this research included the digital images acquired from Agria variety of Ardabil (Iranian northern-west) potato with different sizes and shapes. A combination of morphological features including geometrical features like length, width and features related to shape such as roundness were taken into consideration in identifying irregular potatoes from others employing elongation and Fourier descriptors. Using statistical principal component analysis (PCA), seven features were selected as the most prominent for classification. The experimental results showed that the proposed method achieves a high level of accuracy with merely seven selected discriminative features, obtaining an average correct classification rate of 98% for training set. Additionally, regular potatoes were separated into small, medium and large categories with 100% accuracy. According to the results, the developed algorithm based on image processing can be used in classifying products with no proper shape.

Linear and non-linear modeling to identify vinegars in blends through spectroscopic data

The identification of vinegars produced from six different raw materials (red wine, white wine, cider, apple, molasses, and rice) in blends has been accomplished through their UV–vis spectra and different mathematical models: partial least squares discriminant analysis (PLS-DA) and artificial neural networks (ANNs). The registered spectra were mathematically treated following a linear (PLS-DA) approach and a non-linear one (ANN) based on multilayer perceptron models with different training functions. The average correct classification rate of a series of comparable internal validations was around 55% and 90%, for the PLS-DA and the ANN models respectively, which heavily favors the non-linear approach. Therefore, an accurate chemometric tool with the ability to detect specific vinegars in mixtures in an inexpensive and straightforward fashion has been designed and optimized.

Age group classification and gender detection based on forced expiratory spirometry.

This paper investigates the utility of forced expiratory spirometry (FES) test with efficient machine learning algorithms for the purpose of gender detection and age group classification. The proposed method has three main stages: feature extraction, training of the models and detection. In the first stage, some features are extracted from volume-time curve and expiratory flow-volume loop obtained from FES test. In the second stage, the probabilistic models for each gender and age group are constructed by training Gaussian mixture models (GMMs) and Support vector machine (SVM) algorithm. In the final stage, the gender (or age group) of test subject is estimated by using the trained GMM (or SVM) model. Experiments have been evaluated on a large database from 4571 subjects. The experimental results show that average correct classification rate performance of both GMM and SVM methods based on the FES test is more than 99.3 % and 96.8 % for gender and age group classification, respectively.

Spectra Classification Based on Local Mean-Based K-Nearest Centroid Neighbor Method

In the present paper, a local mean-based K-nearest centroid neighbor (LMKNCN) technique is used for the classification of stars, galaxies and quasars (QSOS). The main idea of LMKNCN is that it depends on the principle of the nearest centroid neighborhood(NCN), and selects K centroid neighbors of each class as training samples and then classifies a query pattern into the class with the distance of the local centroid mean vector to the samples . In this paper, KNN, KNCN and LMKNCN were experimentally compared with these three different kinds of spectra data which are from the United States SDSS-DR8. Among these three methods, the rate of correct classification of the LMKNCN algorithm is higher than the other two algorithms or comparable and the average rate of correct classification is higher than the other two algorithms, especially for the identification of quasars. Experiment shows that the results in this work have important significance for studying galaxies, stars and quasars spectra classification.

Measurement of fermentation parameters of Chinese rice wine using Raman spectroscopy combined with linear and non-linear regression methods

Effective fermentation monitoring is a growing need during the manufacture of wine due to the rapid pace of change in the wine industry. Ethanol and reducing sugar are two most important process variables indicating the status of Chinese rice wine (CRW) fermentation process. In this study, the potentials of Raman spectroscopy (RS) as a rapid process analytical technique to monitor the evolution of these two chemical parameters involved in CRW fermentation process and to group samples according to different fermentation stages were investigated. The results demonstrated that compared with the PLS model using all wavelengths of Raman spectra, the prediction precision of model based on the spectral variables selected by competitive adaptive reweighted sampling (Cars) was significantly improved. In addition, nonlinear models outperformed linear models in predicting fermentation parameters. After systemically comparison and discussion, it was found that for both ethanol and glucose, Cars-support vector machine (Cars-SVM) models gave the best results with the highest prediction precisions. Moreover, the results obtained from discriminant partial least squares analysis (DPLS) showed that good performances were obtained with an average correct classification rate of 94.9% for different fermentation stages. The overall results indicated that RS combined with efficient variable selection algorithm and nonlinear regression tool could be utilized as a rapid method to monitor CRW fermentation process.

Modified multiple antibiotic resistance analysis for the nonpoint source tracking of fecal pollution

Although Multiple Antibiotic Resistance Analysis (MARA) has been adopted for the source tracking of bacterial contamination in natural waters, the accuracy and experimental cost of MARA still have the potential to be improved. Therefore, a process of modified MARA using turbidity was developed, and its feasibility and reliability were evaluated for Fecal Streptococci (FS) in this study. The experimental results are as follows. The development of turbidity via the aesculin hydrolysis by FS could occur in 2–3 hours, and the turbidity of the stock solutions and the number of FS had a proportionate relationship (R2 = 0.991). Thus, the modified MARA could exclude several incubation steps as well as reduce experimental errors by inoculating the same amount of the bacteria into the culture media mixed with antibiotics. The Average Rates of Correct Classification (ARCCs) of the established database based on the modified MARA technique for two-way division and three-way division showed much higher ARCCs (89%) than those of previous studies (61~84%), implying that the modified MARA technique had relatively higher reproducibility than conventional MARA. Also, reliability of the modified MARA was verified by source tracking of stream samples that could be predicted and the real source tracking was carried out by applying the modified MARA in the stream (upstream, midstream, and downstream), which showed reasonable prediction results. The results conclusively demonstrated the modified MARA to be more efficient tool/phenotypic method than conventional MARA for source tracking of fecal contamination in water.

Discrimination of Chinese rice wines of different geographical origins by UV-vis spectroscopy and chemometrics

Discrimination of Chinese rice wines from four geographical origins (‘Zhejiang’, ‘Jiangsu’, ‘Shanghai’ and “Fujian”) in China was investigated according to their UV–vis spectra. The UV–vis absorption spectra of 112 samples in the wavelength range of 190–800 nm were collected. Then the absorption data was subjected to principal component analysis to reveal differences between samples and the potential possibility of using multivariate methods to distinguish differences. Classification models were developed by soft independent modelling of class analogy, linear discriminate analysis (LDA), discriminant partial least squares and support vector machine. Seven mathematical pre-treatments were applied to improve the performance of the multivariate classification models. Among the seven pre-treatments, standard normal transformation (SNV) was superior to the other six methods. The results showed that, compared with other models, SNV-treated LDA models achieved better performances with an average correct classification rate of 98.96% in the training set and 100% in the testing set. The results demonstrate that UV spectroscopy combined with chemometric data analysis, as a rapid method to classify Chinese rice wines according to their geographical origins, is feasible. Copyright © 2015 The Institute of Brewing & Distilling

Resampling-based Classification Using Depth for Functional Curves

The depths, which have been used to detect outliers or to extract a representative subset, can be applied to classification. We propose a resampling-based classification method based on the fact that resampling techniques yield a consistent estimator of the distribution of a statistic. The performance of this method was evaluated with eight contaminated models in terms of Correct Classification Rates (CCRs), and the results were compared with other known methods. The proposed method consistently showed higher average CCRs and 4% higher CCR at the maximum compared to other methods. In addition, this method was applied to Berkeley data. The average CCRs were between 0.79 and 0.85.

Potential of visible/near-infrared hyperspectral imaging for rapid detection of freshness in unfrozen and frozen prawns

The potential of visible and near infrared (400–1000nm) hyperspectral imaging as a rapid and non-invasive method was investigated to differentiate freshness of prawns. In both unfrozen and frozen groups (a total of 280 prawns), two different freshness levels were used for classification, respectively. Mean spectral data from the full surface of prawns were extracted automatically as the hyperspectral cubes. Both the first and second derivative spectra were performed for waveform analysis. Successive projections algorithm (SPA) was conducted to select the individual feature wavelengths for classification. Least squares-support vector machine (LS-SVM), adaptive boosting (AdaBoost) algorithm and back-propagation neutral networks (BP-NN) were carried out for classification using the derivative spectrums based on both full wavelengths and selected feature wavelengths. The results demonstrated that SPA–LS-SVM achieved satisfactory average correct classification rate of 98.33% and 95% for prediction samples in unfrozen and frozen groups, respectively. Visualization map of classification of eight prawns (two groups) was also presented. The overall results revealed that hyperspectral imaging technique is promising for freshness classification of prawns rapidly and non-invasively.

Identifying fecal pollution sources using 3M(™) Petrifilm (™) count plates and antibiotic resistance analysis in the Horse Creek Watershed in Aiken County, SC (USA).

Sources of fecal coliform pollution in a small South Carolina (USA) watershed were identified using inexpensive methods and commonly available equipment. Samples from the upper reaches of the watershed were analyzed with 3M(™) Petrifilm(™) count plates. We were able to narrow down the study's focus to one particular tributary, Sand River, that was the major contributor of the coliform pollution (both fecal and total) to a downstream reservoir that is heavily used for recreation purposes. Concentrations of total coliforms ranged from 2,400 to 120,333cfu/100mL, with sharp increases in coliform counts observed in samples taken after rain events. Positive correlations between turbidity and fecal coliform counts suggested a relationship between fecal pollution and stormwater runoff. Antibiotic resistance analysis (ARA) compared antibiotic resistance profiles of fecal coliform isolates from the stream to those of a watershed-specific fecal source library (equine, waterfowl, canines, and untreated sewage). Known fecal source isolates and unknown isolates from the stream were exposed to six antibiotics at three concentrations each. Discriminant analysis grouped known isolates with an overall average rate of correct classification (ARCC) of 84.3%. A total of 401 isolates from the first stream location were classified as equine (45.9%), sewage (39.4%), waterfowl (6.2%), and feline (8.5%). A similar pattern was observed at the second sampling location, with 42.6% equine, 45.2% sewage, 2.8% waterfowl, 0.6% canine, and 8.8% feline. While there were slight weather-dependent differences, the vast majority of the coliform pollution in this stream appeared to be from two sources, equine and sewage. This information will contribute to better land use decisions and further justify implementation of low-impact development practices within this urban watershed.

Image Classification Recognition for Rock Micro-Thin Section Based on Probabilistic Neural Networks

In order to implement the recognition automation of rock section pore images, a method combined K-means clustering with probabilistic neural network is proposed and applied to rock thin section images. Firstly, K-means clustering is used as segmentation algorithm, the rock images are divided into two types and extracted enough features and it is shown good classification recognition effect on testing dataset. Secondly, 100 pieces of rock image section are used as validation dataset, including 20 groups, each group has 5 images and 200 data samplings. Experiments show that the probabilistic neural network can be used as rock texture classifier, the average correct classification rate is around 95.12%, which can meet the practical application needs.

PREDICTION OF UNIVERSITY ADMISSION ACHIEVEMENT WITH DISCRIMINANT MODEL: CASE STUDY OF PHANATPITTAYAKARN SCHOOL, CHONBURI, THAILAND

This study was conducted to classify two achievement groups of uni- versity admission (failure and success). High school students studied in Phanat- pittayakarn School during the academic year 2006-2011 participated. Gender, high school program, each of eight O-NET scores (Thai, Social, English, Mathematics, Science, Hygiene, Art and Work) as well as GPA of each subject (Social, English, Mathematics, Science, Hygiene, Art, Work) and high school GPA were investigated as predictors of university admission achievement. Discriminant analysis was per- formed to build a predictive model for achievement of university admission. The correct classification rate (CCR) is utilized to evaluate a model performance. The result presents the discriminant classification function obtained rather highly accu- rate of the average CCR (80.79%) for prediction of university admission achievement in the validation data set.