Accuracy Error Rate Research Articles

Meta-cognitive Neural Network based Sequential Learning Framework for Text Categorization

A sequential learning framework for text categorization based on Meta-cognitive Neural Network (McNN) is presented in this paper. Initially text documents are pre-processed and represented in the form of Term Document Matrix (TDM). Since the TDM is of high dimension, to reduce it to lower dimension Regularized Locality Preserving Indexing (RLPI) is used. Further, to categorize the text document, Meta-cognitive Neural Network (McNN) classifier is employed. To measure the effectiveness of the proposed framework, various experiments are conducted on standard benchmark Reuters-21578 dataset and used leave one out cross validation technique to assess the performance. The proposed framework performance is investigated against two well known neural network based classifiers: MLP (Multi Layer Perceptron) and RBF-NN (Radial Basis Function-Neural Network). The experimental results reveals that the McNN classifier uses less number of training documents for learning and it has less true error rate than other two neural network classifiers.

AUTHENTICATION BASED IMAGE FORGERY DETECTION USING OPTIMIZED FEATURES IN JPEG IMAGES

With the current development in technology, image manipulation and particularly image splicing has been termed as a regular and established concern. The swift growth in commercial image editing programs and software for instance Adobe Photoshop has significantly elevated the number of forged, doctored and tampered images on a daily basis. This situation has resulted into extreme consequences, minimizing the authenticity and reliability of such images as well as creating untrue beliefs in a wide range of ideal-world relevance. The authenticity of a digital image suffers from severe threats due to the rise of powerful digital image editing tools that easily alter the image contents without leaving any visible traces of such changes. The splicing forgery can be done by copied a one/more region from source image and pasted into a target image to produce a composite image called spliced image. Therefore, this type of forgery is considered very challenging from tampering detection point of view. To make the matter worst some post-processing effects such as blurring, JPEG compression, rotation and scaling maybe introduced in the spliced image. In previous paper, we study found the problems in forgery detection in Jpg images. Robust feature extraction techniques used in block based copy move image forgery detection require a high computation time. There is a need for reduced computation time for detection schemes to be practical for use with large images and in real time environments. Utilization of GPU to compute the processes can highly parallelize the tasks involved to reduce computation time. Current CPU based algorithms that have been designed are not suitable to be directly adopted in a GPU based scheme. It must be determined how a parallel copy move image forgery detection scheme can be designed for use with a GPU. Digital images provide a new way to represent pictures and scenes that only film and a darkroom could supply before. In this research work, we implement the feature extraction using principle component analysis and optimization (ant colony optimization) algorithm to detect the forgery image in JPG images. In optimization approach to classify the features and match the training feature if training and testing features has matching then detect the forgery image in the jpg images. Evaluate the performance parameters PSNR (Peak Signal to Noise Ratio), Error rate Accuracy and compare the existing parameters i.e accuracy. Keywords- Image Forgery, Robust feature extraction, Ant colony Approach and Neural Network

Redefining the Practice of Peer Review Through Intelligent Automation Part 2: Data-Driven Peer Review Selection and Assignment.

In conventional radiology peer review practice, a small number of exams (routinely 5% of the total volume) is randomly selected, which may significantly underestimate the true error rate within a given radiology practice. An alternative and preferable approach would be to create a data-driven model which mathematically quantifies a peer review risk score for each individual exam and uses this data to identify high risk exams and readers, and selectively target these exams for peer review. An analogous model can also be created to assist in the assignment of these peer review cases in keeping with specific priorities of the service provider. An additional option to enhance the peer review process would be to assign the peer review cases in a truly blinded fashion. In addition to eliminating traditional peer review bias, this approach has the potential to better define exam-specific standard of care, particularly when multiple readers participate in the peer review process.

Item-by-item sampling for promotional purposes

ABSTRACT In this paper, we present a method for sampling items that are checked on a pass/fail basis, with a view to a statement being made about the success/failure rate for the purposes of promoting an organisation’s product/service to potential clients/customers. Attention is paid to the appropriate use of statistical phrases for the statements and this leads to the use of Bayesian credible intervals, thus it exceeds what can be achieved with standard acceptance sampling techniques. The hypergeometric distribution is used to calculate successive stopping rules so that the resources used for sampling can be minimised. Extensions to the sampling procedure are considered to allow the potential for stronger and weaker statements to be made as sampling progresses. The relationship between the true error rate and the probabilities of making correct statements is discussed.

Effect of Correlations on Type 1 Error Rates of Some Multivariate Normality Tests

Normality assumption of multivariate data is a prerequisite to the use of multivariate statistical data analysis methods before inference could be valid and reliable. Tests developed to validate this assumption including Doornik-Harsen (DH), Shapiro-Francia (SF), Mardia Skewness (MS), Mardia Skewness for small sample (MSS) and Kurtosis (MK), Skewness (S) and Kurtosis(K), Shapiro-Wilk(SW), Royston (R), Desgagne-Micheaux (DM), Henze-Zirkler (HZ), Energy (E), Gel-Gastwirth (GG) and Bontemps-Meddahi (BM) tests often result into different conclusions. These differences can be misleading. Consequently, this paper examined the effect of correlations on the Type 1 error rates of multivariate tests of normality. Monte Carlo experiments were conducted one thousand (1000) times taking into consideration the dimensions, correlations and sample sizes of the multivariate data. A test is affected by correlation if its estimated Type 1 error rate changes as correlation changes. A test is considered good if its estimated error rate approximates the true error rate and best if the number of times it approximates the estimated error rate when counted over the levels of correlations, sample sizes and levels of significance is the highest, the mode. Results show that Type 1 error rates of DH, SF, SW, R, DM, GG and BM tests are affected by correlations and are relatively not good; where as the Type 1 error rates of HZ, MS, MK, MSS, S, K and E tests are not only unaffected by correlations but are also relatively good. Consequently, MS, R, MSS, HZ and E tests have good Type 1 error rates but that of E and HZ tests are best. They are therefore recommended for practitioners.

SEUs Mitigation on Program Counter of the LEON3 Soft Processor

Analyzing and evaluating the sensitivity of embedded systems to soft-errors have always been a challenge for aerospace or safety equipment designer. Different automated fault-injection methods have been developed for evaluating the sensitivity of integrated circuit. Also many techniques have been developed to get a fault tolerant architecture in order to mask and mitigate fault injection in a circuit. Fault injection mitigation and repair techniques are applied together on LEON3 processor in goal to study the reliability of a soft-core. The so-called NETlist Fault Injection (NETFI+) tool is a fault injection techniques used in this paper. The prediction of Single Event Upset (SEU) error-rates between radiation ground testing and FPGA implementation have been done with good and accurate result. But no functional simulations have been performed. A Triple Modular Redundancy (TMR) is used in this paper as a repair technique versus fault injection. This paper analyses the effectiveness of fault tolerant method on LEON3 soft-core running a benchmark. It starts by evaluating the behavior of LEON3’s program counter against Single Event Upset error-rate accuracy between the functional simulation and the FPGA emulation and an analysis of the LEON3 reliability in presence of fault tolerant technique. The objective is to offer, through the new version of NETFI+ with introducing a fault tolerant technique, the possibility to designers to evaluate the benefits of SEUs mitigation for the LEON3 processor on the program counter.

Comparative Analysis of Selected Methods for Estimating the Prediction Error of Classifie

Classification is an algorithm, which assigns studied companies, taking into consideration their attributes, to specific population. An essential part of it is classifier. Its measure of quality is especially predictability, measured by true error rate. The value of this error, due to lack of sufficiently large and independent test set, must be estimated on the basis of available learning set.The aim of this article is to make a review and compare selected methods for estimating the prediction error of classifier, constructed with linear discriminant analysis. It was examined if the results of the analysis depends on the sample size and the method of selecting variables for a model. Empirical research was made on example of problem of bankruptcy prediction of join-stock companies in Poland.

Empirical Investigation of Type 1 Error Rate of Univariate Tests of Normality

Normality assumption is important in univariate parametric statistical tests. Either the varibles or the error terms in the model have to be normally distributed before valid statistical conclusions could be made. Various tests of univariate normality including that of Pearson, Kolmogorov–Smirnov, Anderson-Darling, Shapiro–Wilk, Lilliefor, D’Agostino and Pearson, Jarque-Bera, Shapiro-Franca, Energy and Cramervon Mises tests have been developed. However, when applied in practice, they hardly give the same result. Thus, this research work aims at investigating the Type 1 error rate of these tests so as to identify the best one and suggest the same for statistics users. The tests were compared by conducting Monte Carlo experiments five thousand (5000) times with six sample sizes at three pre-selected levels of significance. A test was adjudged good at a particular level of significance if its empirical Type 1 error rate approximated the true error rates most often. It is best if its number of counts at which it was good over the sample sizes and levels of significance was the highest. Results reveal that Type 1 error rate of all the univariate tests are good except that of Kolmogorov–Smirnov, Pearson Unadjusted and Jarque-Bera. Moreover, those of Anderson-Darling, Shapiro-Wilk, Energy and Cramer-von Mises tests are relatively best. They are therefore recommended for testing the assumption of normality in any univariate data set.

Bounding quantum gate error rate based on reported average fidelity

Remarkable experimental advances in quantum computing are exemplified by recent announcements of impressive average gate fidelities exceeding 99.9% for single-qubit gates and 99% for two-qubit gates. Although these high numbers engender optimism that fault-tolerant quantum computing is within reach, the connection of average gate fidelity with fault-tolerance requirements is not direct. Here we use reported average gate fidelity to determine an upper bound on the quantum-gate error rate, which is the appropriate metric for assessing progress towards fault-tolerant quantum computation, and we demonstrate that this bound is asymptotically tight for general noise. Although this bound is unlikely to be saturated by experimental noise, we demonstrate using explicit examples that the bound indicates a realistic deviation between the true error rate and the reported average fidelity. We introduce the Pauli distance as a measure of this deviation, and we show that knowledge of the Pauli distance enables tighter estimates of the error rate of quantum gates.

Technical errors and complications in orthopaedic trauma surgery.

IntroductionAdverse events and associated morbidity and subsequent costs receive increasing attention in clinical practice and research. As opposed to complications, errors are not described or analysed in literature on fracture surgery. The aim of this study was to provide a description of errors and complications in relation to fracture surgery, as well as the circumstances in which they occur, for example urgency, type of surgeon, and type of fracture.MethodsAll errors and complications were recorded prospectively in our hospital’s complication registry, which forms an integral part of the electronic medical patient file. All recorded errors and complications in the complication registry linked to fracture surgery between 1 January, 2000 and 31 December, 2010 were analysed.ResultsDuring the study period 4310 osteosynthesis procedures were performed. In 78 (1.8 %) procedures an error in osteosynthesis was registered. The number of procedures in which an error occurred was significantly lower (OR = 0.53; p = 0.007) when an orthopaedic trauma surgeon was part of the operating team. Of all 3758 patients who were admitted to the surgical ward for osteosynthesis, 745 (19.8 %) had one or more postoperative complications registered. There was no significant difference in the number of postoperative complications after osteosynthesis procedures in which an orthopaedic trauma surgeon was present or absent (16.7 vs. 19.1 %; p = 0.088; OR 0.85).DiscussionIn the present study the true error rate after osteosynthesis may have been higher than the rate found. Errors that had no significant consequence may be especially susceptible to underreporting.ConclusionThe present study suggests that an osteosynthesis procedure performed by or actively assisted by an orthopaedic trauma surgeon decreases the probability of an error in osteosynthesis. Apart from errors in osteosynthesis, the involvement of an orthopaedic trauma surgeon did not lead to a significant reduction in the number of postoperative complications.

Generalized Consistent Error Estimator of Linear Discriminant Analysis

A classifier is epistemologically vacuous without an accurate estimate of its true error rate. In situations where the number of sample points is of the same order of magnitude as the dimension of observations, serious issues arise with respect to the performance of error estimators. In this paper, we place the problem of synthesizing an error rate estimator of a common linear classifier in an asymptotic setting in which the number of sample points is kept comparable in magnitude to the dimension of observations (double asymptotic). We construct a generalized consistent estimator of the true error rate for linear discriminant analysis in the multivariate Gaussian model under the assumption of a common covariance matrix. In other words, the estimator converges to true error rate in the double asymptotic sense. We employ simulations using both synthetic and real data to compare the performance of the new estimator to the classical consistent estimator of the true error (plug-in estimator) as well as other well-known estimators. We observe that the constructed estimator can outperform other estimators of the true error in many situations in terms of bias and root-mean-square (RMS) error.

Correcting for Sequencing Error in Maximum Likelihood Phylogeny Inference

Accurate phylogenies are critical to taxonomy as well as studies of speciation processes and other evolutionary patterns. Accurate branch lengths in phylogenies are critical for dating and rate measurements. Such accuracy may be jeopardized by unacknowledged sequencing error. We use simulated data to test a correction for DNA sequencing error in maximum likelihood phylogeny inference. Over a wide range of data polymorphism and true error rate, we found that correcting for sequencing error improves recovery of the branch lengths, even if the assumed error rate is up to twice the true error rate. Low error rates have little effect on recovery of the topology. When error is high, correction improves topological inference; however, when error is extremely high, using an assumed error rate greater than the true error rate leads to poor recovery of both topology and branch lengths. The error correction approach tested here was proposed in 2004 but has not been widely used, perhaps because researchers do not want to commit to an estimate of the error rate. This study shows that correction with an approximate error rate is generally preferable to ignoring the issue.

Randomly stopped sums: models and psychological applications.

This paper describes an approach to modeling the sums of a continuous random variable over a number of measurement occasions when the number of occasions also is a random variable. A typical example is summing the amounts of time spent attending to pieces of information in an information search task leading to a decision to obtain the total time taken to decide. Although there is a large literature on randomly stopped sums in financial statistics, it is largely absent from psychology. The paper begins with the standard modeling approaches used in financial statistics, and then extends them in two ways. First, the randomly stopped sums are modeled as “life distributions” such as the gamma or log-normal distribution. A simulation study investigates Type I error rate accuracy and power for gamma and log-normal versions of this model. Second, a Bayesian hierarchical approach is used for constructing an appropriate general linear model of the sums. Model diagnostics are discussed, and three illustrations are presented from real datasets.

Assessment of competency in clinical measurement: comparison of two forms of sequential test and sensitivity of test error rates to parameter choice

To assess clinical measurement competency by two sequential test formulations [resetting sequential probability ratio test (R-SPRT) and learning curve cumulative summation (LC-CUSUM)]. Numerical simulation and retrospective observational study. Obstetric ultrasound department. Cohorts of 10,000 simulated trainees and 62 obstetric sonographers training in nuchal translucency (NT) measurement at the 11-14-week pregnancy scan with limited case availability. Application of LC-CUSUM and R-SPRT to clinical measurement training. Proportions of real trainees achieving competency by LC-CUSUM and R-SPRT, proportions of simulated competent trainees not achieving competency (Type I error), proportions of simulated incompetent trainees achieving competency (Type II error), distribution of case number required to achieve competency (run length) and frequency of resets. For simulated cohorts, significant differences in run-length distribution and true test error rates were found between the R-SPRT and LC-CUSUM tests with equivalent parameters. Increasing the cases available to each trainee reduced the Type I error rate but increased the Type II error rate for both sequential tests for all choices of unacceptable failure rate. Discontinuities in the proportion of trainees expected to be test competent were found at critical values of unacceptable failure rate. With equivalent parameters, the R-SPRT and LC-CUSUM formulations of sequential tests produced different outcomes, demonstrating that the choice of test method, as well as the choice of parameters, is important in designing a training scheme. The R-SPRT detects incompetence as well as competence and may indicate need for further training. Simulations are valuable in estimating the proportions of trainees expected to be assessed as competent.

Classifier design given an uncertainty class of feature distributions via regularized maximum likelihood and the incorporation of biological pathway knowledge in steady-state phenotype classification

Contemporary high-throughput technologies provide measurements of very large numbers of variables but often with very small sample sizes. This paper proposes an optimization-based paradigm for utilizing prior knowledge to design better performing classifiers when sample sizes are limited. We derive approximate expressions for the first and second moments of the true error rate of the proposed classifier under the assumption of two widely used models for the uncertainty classes: ε-contamination and p-point classes. The applicability of the approximate expressions is discussed by defining the problem of finding optimal regularization parameters through minimizing the expected true error. Simulation results using the Zipf model show that the proposed paradigm yields improved classifiers that outperform traditional classifiers which use only training data. Our application of interest involves discrete gene regulatory networks possessing labeled steady-state distributions. Given prior operational knowledge of the process, our goal is to build a classifier that can accurately label future observations obtained in the steady state by utilizing both the available prior knowledge and the training data. We examine the proposed paradigm on networks containing NF-κB pathways, where it shows significant improvement in classifier performance over the classical data-only approach to classifier design. Companion website: http://gsp.tamu.edu/Publications/supplementary/shahrokh12a.

Fuzzy logic method for motor quality types on current waveforms

Fuzzy logic theory is widely adopted in various fields, such as speech recognition, image retrieval and pattern recognition. This study proposes a fuzzy logic approach to analyze current waveform for determining the motor’s quality types. Proper qualitative features of motor’s current waveforms are selected. Based on selected qualitative features, motor’s quality types are determined by using the fuzzy logic method. The accuracy of error rate is 7.377% (0.294%) for wrong judgment on defect (good) motor to be determined as good (defect). The right rate is 92.622% (99.705%) for right judgment on defect (good) motor to be determined as defect (good).

Performance and estimation of the true error rate of classification rules built with additional information. An application to a cancer trial

Classification rules that incorporate additional information usually present in discrimination problems are receiving certain attention during the last years as they perform better than the usual rules. Fernández, M. A., C. Rueda and B. Salvador (2006): "Incorporating additional information to normal linear discriminant rules," J. Am. Stat. Assoc., 101, 569-577, proved that these rules have lower total misclassification probability than the usual Fisher's rule. In this paper we consider two issues; on the one hand, we compare these rules with those based on shrinkage estimators of the mean proposed by Tong, T., L. Chen and H. Zhao (2012): "Improved mean estimation and its application to diagonal discriminant analysis," Bioinformatics, 28(4): 531-537. with regard to four criteria: total misclassification probability, area under ROC curve, well-calibratedness and refinement; on the other hand, we consider the estimation of the true error rate, which is a very interesting parameter in applications. We prove results on the apparent error rate of the rules that expose the need of new estimators of their true error rate. We propose four such new estimators. Two of them are defined incorporating the additional information into the leave-one-out-bootstrap. The other two are the corresponding cross-validation after bootstrap versions. We compare these estimators with the usual ones in a simulation study and in a cancer trial application, showing the good behavior of the rules that incorporate additional information and of the new leave-one-out bootstrap estimators of their true error rate.

Estimation of genotyping error rate from repeat genotyping, unintentional recaptures and known parent–offspring comparisons in 16 microsatellite loci for brown rockfish (Sebastes auriculatus)

Genotyping errors are present in almost all genetic data and can affect biological conclusions of a study, particularly for studies based on individual identification and parentage. Many statistical approaches can incorporate genotyping errors, but usually need accurate estimates of error rates. Here, we used a new microsatellite data set developed for brown rockfish (Sebastes auriculatus) to estimate genotyping error using three approaches: (i) repeat genotyping 5% of samples, (ii) comparing unintentionally recaptured individuals and (iii) Mendelian inheritance error checking for known parent-offspring pairs. In each data set, we quantified genotyping error rate per allele due to allele drop-out and false alleles. Genotyping error rate per locus revealed an average overall genotyping error rate by direct count of 0.3%, 1.5% and 1.7% (0.002, 0.007 and 0.008 per allele error rate) from replicate genotypes, known parent-offspring pairs and unintentionally recaptured individuals, respectively. By direct-count error estimates, the recapture and known parent-offspring data sets revealed an error rate four times greater than estimated using repeat genotypes. There was no evidence of correlation between error rates and locus variability for all three data sets, and errors appeared to occur randomly over loci in the repeat genotypes, but not in recaptures and parent-offspring comparisons. Furthermore, there was no correlation in locus-specific error rates between any two of the three data sets. Our data suggest that repeat genotyping may underestimate true error rates and may not estimate locus-specific error rates accurately. We therefore suggest using methods for error estimation that correspond to the overall aim of the study (e.g. known parent-offspring comparisons in parentage studies).

Latent class analysis of response inconsistencies across modes of data collection

Latent class analysis (LCA) has been hailed as a promising technique for studying measurement errors in surveys, because the models produce estimates of the error rates associated with a given question. Still, the issue arises as to how accurate these error estimates are and under what circumstances they can be relied on. Skeptics argue that latent class models can understate the true error rates and at least one paper (Kreuter et al., 2008) demonstrates such underestimation empirically. We applied latent class models to data from two waves of the National Survey of Family Growth (NSFG), focusing on a pair of similar items about abortion that are administered under different modes of data collection. The first item is administered by computer-assisted personal interviewing (CAPI); the second, by audio computer-assisted self-interviewing (ACASI). Evidence shows that abortions are underreported in the NSFG and the conventional wisdom is that ACASI item yields fewer false negatives than the CAPI item. To evaluate these items, we made assumptions about the error rates within various subgroups of the population; these assumptions were needed to achieve an identifiable LCA model. Because there are external data available on the actual prevalence of abortion (by subgroup), we were able to form subgroups for which the identifying restrictions were likely to be (approximately) met and other subgroups for which the assumptions were likely to be violated. We also ran more complex models that took potential heterogeneity within subgroups into account. Most of the models yielded implausibly low error rates, supporting the argument that, under specific conditions, LCA models underestimate the error rates.

Does parametric fMRI analysis with SPM yield valid results?—An empirical study of 1484 rest datasets

The validity of parametric functional magnetic resonance imaging (fMRI) analysis has only been reported for simulated data. Recent advances in computer science and data sharing make it possible to analyze large amounts of real fMRI data. In this study, 1484 rest datasets have been analyzed in SPM8, to estimate true familywise error rates. For a familywise significance threshold of 5%, significant activity was found in 1%-70% of the 1484 rest datasets, depending on repetition time, paradigm and parameter settings. This means that parametric significance thresholds in SPM both can be conservative or very liberal. The main reason for the high familywise error rates seems to be that the global AR(1) auto correlation correction in SPM fails to model the spectra of the residuals, especially for short repetition times. The findings that are reported in this study cannot be generalized to parametric fMRI analysis in general, other software packages may give different results. By using the computational power of the graphics processing unit (GPU), the 1484 rest datasets were also analyzed with a random permutation test. Significant activity was then found in 1%-19% of the datasets. These findings speak to the need for a better model of temporal correlations in fMRI timeseries.