A kernel-based two-class classifier for imbalanced data sets

Many kernel classifier construction algorithms adopt classification accuracy as performance metrics in model evaluation. Moreover, equal weighting is often applied to each data sample in parameter estimation. These modeling practices often become problematic if the data sets are imbalanced. We present a kernel classifier construction algorithm using orthogonal forward selection (OFS) in order to optimize the model generalization for imbalanced two-class data sets. This kernel classifier identification algorithm is based on a new regularized orthogonal weighted least squares (ROWLS) estimator and the model selection criterion of maximal leave-one-out area under curve (LOO-AUC) of the receiver operating characteristics (ROCs). It is shown that, owing to the orthogonalization procedure, the LOO-AUC can be calculated via an analytic formula based on the new regularized orthogonal weighted least squares parameter estimator, without actually splitting the estimation data set. The proposed algorithm can achieve minimal computational expense via a set of forward recursive updating formula in searching model terms with maximal incremental LOO-AUC value. Numerical examples are used to demonstrate the efficacy of the algorithm.

Blood-clotting response tests for resistance to diphacinone and chlorophacinone in the Norway rat (Rattus norvegicus Berk.)

Resistance baselines were obtained for the first generation anticoagulant rodenticides chlorophacinone and diphacinone using laboratory, caesarian-derived Norway rats (Rattus norvegicus) as the susceptible strain and the blood clotting response test method. The ED99 estimates for a quantal response were: chlorophacinone, males 0.86 mg kg−1, females 1.03 mg kg−1; diphacinone, males 1.26 mg kg−1, females 1.60 mg kg−1. The dose-response data also showed that chlorophacinone was significantly (p<0.0001) more potent than diphacinone for both male and female rats, and that male rats were more susceptible than females to both compounds (p<0.002). The ED99 doses were then given to groups of five male and five female rats of the Welsh and Hampshire warfarin-resistant strains. Twenty-four hours later, prothrombin times were slightly elevated in both strains but all the animals were classified as resistant to the two compounds, indicating cross-resistance from warfarin to diphacinone and chlorophacinone. When rats of the two resistant strains were fed for six consecutive days on baits containing either diphacinone or chlorophacinone, many animals survived, indicating that their resistance might enable them to survive treatments with these compounds in the field.

The persistence of real estate fund performance

This paper examines the short and long-term persistence of tax-exempt real estate funds in the UK through the use of winner-loser contingency table methodology. The persistence tests are applied to a database of varying numbers of funds from a low of 16 to a high of 27 using quarterly returns over the 12 years from 1990 Q1 to 2001 Q4. The overall conclusion is that the real estate funds in the UK show little evidence of persistence in the short-term (quarterly and semi-annual data) or for data over a considerable length of time (bi-annual to six yearly intervals). In contrast, the results are better for annual data with evidence of significant performance persistence. Thus at this stage, it seems that an annual evaluation period, provides the best discrimination of the winner and loser phenomenon in the real estate market. This result is different from equity and bond studies, where it seems that the repeat winner phenomenon is stronger over shorter periods of evaluation. These results require careful interpretation, however, as the results show that when only small samples are used significant adjustments must be made to correct for small sample bias and second the conclusions are sensitive to the length of the evaluation period and specific test used. Nonetheless, it seems that persistence in performance of real estate funds in the UK does exist, at least for the annual data, and it appears to be a guide to beating the pack in the long run. Furthermore, although the evidence of persistence in performance for the overall sample of funds is limited, we have found evidence that two funds were consistent winners over this period, whereas no one fund could be said to be a consistent loser.

A sequential two meal challenge reveals abnormalities in postprandial TAG but not glucose in men with increasing numbers of metabolic syndrome components

Objective To examine the impact of increasing numbers of metabolic syndrome (MetS) components on postprandial lipaemia. Methods Healthy men (n = 112) underwent a sequential meal postprandial investigation, in which blood samples were taken at regular intervals after a test breakfast (0 min) and lunch (330 min). Lipids and glucose were measured in the fasting sample, with triacylglycerol (TAG), non-esterified fatty acids and glucose analysed in the postprandial samples. Results Subjects were grouped according to the number of MetS components regardless of the combinations of components (0/1, 2, 3 and 4/5). As expected, there was a trend for an increase in body mass index, blood pressure, fasting TAG, glucose and insulin, and a decrease in fasting high-density lipoprotein cholesterol with increasing numbers of MetS components (P≤0.0004). A similar trend was observed for the summary measures of the postprandial TAG and glucose responses. For TAG, the area under the curve (AUC) and maximum concentration (maxC) were significantly greater in men with ≥ 3 than < 3 components (P < 0.001), whereas incremental AUC was greater in those with 3 than 0/1 and 2, and 4/5 compared with 2 components (P < 0.04). For glucose, maxC after the test breakfast (0-330 min) and total AUC (0-480 min) were higher in men with ≥ 3 than < 3 components (P≤0.001). Conclusions Our data analysis has revealed a linear trend between increasing numbers of MetS components and magnitude (AUC) of the postprandial TAG and glucose responses. Furthermore, the two meal challenge discriminated a worsening of postprandial lipaemic control in subjects with ≥ 3 MetS components.

Deception-detection and machine intelligence in practical Turing tests

Deception-detection is the crux of Turing’s experiment to examine machine thinking conveyed through a capacity to respond with sustained and satisfactory answers to unrestricted questions put by a human interrogator. However, in 60 years to the month since the publication of Computing Machinery and Intelligence little agreement exists for a canonical format for Turing’s textual game of imitation, deception and machine intelligence. This research raises from the trapped mine of philosophical claims, counter-claims and rebuttals Turing’s own distinct five minutes question-answer imitation game, which he envisioned practicalised in two different ways: a) A two-participant, interrogator-witness viva voce, b) A three-participant, comparison of a machine with a human both questioned simultaneously by a human interrogator. Using Loebner’s 18th Prize for Artificial Intelligence contest, and Colby et al.’s 1972 transcript analysis paradigm, this research practicalised Turing’s imitation game with over 400 human participants and 13 machines across three original experiments. Results show that, at the current state of technology, a deception rate of 8.33% was achieved by machines in 60 human-machine simultaneous comparison tests. Results also show more than 1 in 3 Reviewers succumbed to hidden interlocutor misidentification after reading transcripts from experiment 2. Deception-detection is essential to uncover the increasing number of malfeasant programmes, such as CyberLover, developed to steal identity and financially defraud users in chatrooms across the Internet. Practicalising Turing’s two tests can assist in understanding natural dialogue and mitigate the risk from cybercrime.

Measuring perceived brand luxury: an evaluation of the BLI scale

This paper investigates the psychometric properties of Vigneron and Johnson's Brand Luxury Index scale. The authors developed the scale using data collected from a student sample in Australia. To validate the scale, the study reported in this paper uses data collected from Taiwanese luxury consumers. The scale was initially subjected to reliability analysis yielding low α values for two of its five proposed dimensions. Exploratory and confirmatory factors analyses were subsequently performed to examine the dimensionality of brand luxury. Discriminant and convergent validity tests highlight the need for further research into the dimensionality of the construct. Although the scale represents a good initial contribution to understanding brand luxury, in view of consumers' emerging shopping patterns, further investigation is warranted to establish the psychometric properties of the scale and its equivalence across cultures.

A two-dimensional X-ray scattering system for in-situ time-resolving studies of polymer structures subjected to controlled deformations

A two-dimensional X-ray scattering system developed around a CCD-based area detector is presented, both in terms of hardware employed and software designed and developed. An essential feature is the integration of hardware and software, detection and sample environment control which enables time-resolving in-situ wide-angle X-ray scattering measurements of global structural and orientational parameters of polymeric systems subjected to a variety of controlled external fields. The development and operation of a number of rheometers purpose-built for the application of such fields are described. Examples of the use of this system in monitoring degrees of shear-induced orientation in liquid-crystalline systems and crystallization of linear polymers subsequent to shear flow are presented.

Reduced cortico-motor facilitation in a normal sample with high traits of autism

Recent research in social neuroscience proposes a link between mirror neuron system (MNS) and social cognition. The MNS has been proposed to be the neural mechanism underlying action recognition and intention understanding and more broadly social cognition. Pre-motor MNS has been suggested to modulate the motor cortex during action observation. This modulation results in an enhanced cortico-motor excitability reflected in increased motor evoked potentials (MEPs) at the muscle of interest during action observation. Anomalous MNS activity has been reported in the autistic population whose social skills are notably impaired. It is still an open question whether traits of autism in the normal population are linked to the MNS functioning. We measured TMS-induced MEPs in normal individuals with high and low traits of autism as measured by the autistic quotient (AQ), while observing videos of hand or mouth actions, static images of a hand or mouth or a blank screen. No differences were observed between the two while they observed a blank screen. However participants with low traits of autism showed significantly greater MEP amplitudes during observation of hand/mouth actions relative to static hand/mouth stimuli. In contrast, participants with high traits of autism did not show such a MEP amplitude difference between observation of actions and static stimuli. These results are discussed with reference to MNS functioning.

A practical comparison of blinded methods for sample size reviews in survival data clinical trials.

This paper presents practical approaches to the problem of sample size re-estimation in the case of clinical trials with survival data when proportional hazards can be assumed. When data are readily available at the time of the review, on a full range of survival experiences across the recruited patients, it is shown that, as expected, performing a blinded re-estimation procedure is straightforward and can help to maintain the trial's pre-specified error rates. Two alternative methods for dealing with the situation where limited survival experiences are available at the time of the sample size review are then presented and compared. In this instance, extrapolation is required in order to undertake the sample size re-estimation. Worked examples, together with results from a simulation study are described. It is concluded that, as in the standard case, use of either extrapolation approach successfully protects the trial error rates. Copyright © 2012 John Wiley & Sons, Ltd.

Sensitivity and out-of-sample error in continuous time data assimilation

Data assimilation refers to the problem of finding trajectories of a prescribed dynamical model in such a way that the output of the model (usually some function of the model states) follows a given time series of observations. Typically though, these two requirements cannot both be met at the same time–tracking the observations is not possible without the trajectory deviating from the proposed model equations, while adherence to the model requires deviations from the observations. Thus, data assimilation faces a trade-off. In this contribution, the sensitivity of the data assimilation with respect to perturbations in the observations is identified as the parameter which controls the trade-off. A relation between the sensitivity and the out-of-sample error is established, which allows the latter to be calculated under operational conditions. A minimum out-of-sample error is proposed as a criterion to set an appropriate sensitivity and to settle the discussed trade-off. Two approaches to data assimilation are considered, namely variational data assimilation and Newtonian nudging, also known as synchronization. Numerical examples demonstrate the feasibility of the approach.

An effective approach for land-cover classification from airborne lidar fused with co-registered data

Airborne lidar provides accurate height information of objects on the earth and has been recognized as a reliable and accurate surveying tool in many applications. In particular, lidar data offer vital and significant features for urban land-cover classification, which is an important task in urban land-use studies. In this article, we present an effective approach in which lidar data fused with its co-registered images (i.e. aerial colour images containing red, green and blue (RGB) bands and near-infrared (NIR) images) and other derived features are used effectively for accurate urban land-cover classification. The proposed approach begins with an initial classification performed by the Dempster–Shafer theory of evidence with a specifically designed basic probability assignment function. It outputs two results, i.e. the initial classification and pseudo-training samples, which are selected automatically according to the combined probability masses. Second, a support vector machine (SVM)-based probability estimator is adopted to compute the class conditional probability (CCP) for each pixel from the pseudo-training samples. Finally, a Markov random field (MRF) model is established to combine spatial contextual information into the classification. In this stage, the initial classification result and the CCP are exploited. An efficient belief propagation (EBP) algorithm is developed to search for the global minimum-energy solution for the maximum a posteriori (MAP)-MRF framework in which three techniques are developed to speed up the standard belief propagation (BP) algorithm. Lidar and its co-registered data acquired by Toposys Falcon II are used in performance tests. The experimental results prove that fusing the height data and optical images is particularly suited for urban land-cover classification. There is no training sample needed in the proposed approach, and the computational cost is relatively low. An average classification accuracy of 93.63% is achieved.

Knowledge of A/A'-dependencies on subject extraction with two types of infinitives in non-native Portuguese adult bilingualism

Within generative L2 acquisition research there is a longstanding debate as to what underlies observable differences in L1/L2 knowledge/ performance. On the one hand, Full Accessibility approaches maintain that target L2 syntactic representations (new functional categories and features) are acquirable (e.g., Schwartz & Sprouse, 1996). Conversely, Partial Accessibility approaches claim that L2 variability and/or optionality, even at advanced levels, obtains as a result of inevitable deficits in L2 narrow syntax and is conditioned upon a maturational failure in adulthood to acquire (some) new functional features (e.g., Beck, 1998; Hawkins & Chan, 1997; Hawkins & Hattori, 2006; Tsimpli & Dimitrakopoulou, 2007). The present study tests the predictions of these two sets of approaches with advanced English learners of L2 Brazilian Portuguese (n = 21) in the domain of inflected infinitives. These advanced L2 learners reliably differentiate syntactically between finite verbs, uninflected and inflected infinitives, which, as argued, only supports Full Accessibility approaches. Moreover, we will discuss how testing the domain of inflected infinitives is especially interesting in light of recent proposals that Brazilian Portuguese colloquial dialects no longer actively instantiate them (Lightfoot, 1991; Pires, 2002, 2006; Pires & Rothman, 2009; Rothman, 2007).

Forecast encompassing tests and probability forecasts

We consider tests of forecast encompassing for probability forecasts, for both quadratic and logarithmic scoring rules. We propose test statistics for the null of forecast encompassing, present the limiting distributions of the test statistics, and investigate the impact of estimating the forecasting models' parameters on these distributions. The small-sample performance is investigated, in terms of small numbers of forecasts and model estimation sample sizes. We show the usefulness of the tests for the evaluation of recession probability forecasts from logit models with different leading indicators as explanatory variables, and for evaluating survey-based probability forecasts.

Evaluating multivariate forecast densities: a comparison of two approaches

We consider methods of evaluating multivariate density forecasts. A recently proposed method is found to lack power when the correlation structure is mis-specified. Tests that have good power to detect mis-specifications of this sort are described. We also consider the properties of the tests in the presence of more general mis-specifications.

A comparison of tests of non-linear cointegration with an application to the predictability of US interest rates using the term structure

We test whether there are nonlinearities in the response of short- and long-term interest rates to the spread in interest rates, and assess the out-of-sample predictability of interest rates using linear and nonlinear models. We find strong evidence of nonlinearities in the response of interest rates to the spread. Nonlinearities are shown to result in more accurate short-horizon forecasts, especially of the spread.