Search results

The purpose of this paper is to assess the performance of competing methods and model selection, which are non‐trivial issues given the financial implications. Researchers have adopted various methods including statistical models and machine learning methods such as neural networks to assist decision making in direct marketing. However, due to the different performance criteria and validation techniques currently in practice, comparing different methods is often not straightforward.

This study compares the performance of neural networks with that of classification and regression tree, latent class models and logistic regression using three criteria – simple error rate, area under the receiver operating characteristic curve (AUROC), and cumulative lift – and two validation methods, i.e. bootstrap and stratified k‐fold cross‐validation. Systematic experiments are conducted to compare their performance.

The results suggest that these methods vary in performance across different criteria and validation methods. Overall, neural networks outperform the others in AUROC value and cumulative lifts, and the stratified ten‐fold cross‐validation produces more accurate results than bootstrap validation.

To select predictive models to support direct marketing decisions, researchers need to adopt appropriate performance criteria and validation procedures.

The study addresses the key issues in model selection, i.e. performance criteria and validation methods, and conducts systematic analyses to generate the findings and practical implications.

Based on user-generated content from a Chinese social media platform, this paper aims to investigate multiple methods of constructing user profiles and their effectiveness in predicting their gender, age and geographic location.

This investigation collected 331,634 posts from 4,440 users of Sina Weibo. The data were divided into two parts, for training and testing . First, a vector space model and topic models were applied to construct user profiles. A classification model was then learned by a support vector machine according to the training data set. Finally, we used the classification model to predict users’ gender, age and geographic location in the testing data set.

The results revealed that in constructing user profiles, latent semantic analysis performed better on the task of predicting gender and age. By contrast, the method based on a traditional vector space model worked better in making predictions regarding the geographic location. In the process of applying a topic model to construct user profiles, the authors found that different prediction tasks should use different numbers of topics.

This study explores different user profile construction methods to predict Chinese social media network users’ gender, age and geographic location. The results of this paper will help to improve the quality of personal information gathered from social media platforms, and thereby improve personalized recommendation systems and personalized marketing.

The purpose of this paper is to compare various linear opinion pooling models for aggregating probability judgments and to determine whether Cooke's performance weighting model can sift out better calibrated experts and produce better aggregated distribution.

The leave‐one‐out cross‐validation technique is adopted to perform an out‐of‐sample comparison of Cooke's classical model, the equal weight linear pooling method, and the best expert approach.

Both aggregation models significantly outperform the best expert approach, indicating the need for inputs from multiple experts. The performance score for Cooke's classical model drops considerably in out‐of‐sample analysis, indicating that Cooke's performance weight approach might have been slightly overrated before, and the performance weight aggregation method no longer dominantly outperforms the equal weight linear opinion pool.

The results show that using seed questions to sift out better calibrated experts may still be a feasible approach. However, because the superiority of Cooke's model as discussed in previous studies can no longer be claimed, whether the cost of extra efforts used in generating and evaluating seed questions is justifiable remains a question.

Understanding the performance of various models for aggregating experts' probability judgments is critical for decision and risk analysis. Furthermore, the leave‐one‐out cross‐validation technique used in this study achieves more objective evaluations than previous studies.

Social signal processing under affective computing aims at recognizing and extracting useful human social interaction patterns. Fight is a common social interaction in real life. A fight detection system finds wide applications. This paper aims to detect fights in a natural and low-cost manner.

Research works on fight detection are often based on visual features, demanding substantive computation and good video quality. In this paper, the authors propose an approach to detect fight events through motion analysis. Most existing works evaluated their algorithms on public data sets manifesting simulated fights, where the fights are acted out by actors. To evaluate real fights, the authors collected videos involving real fights to form a data set. Based on the two types of data sets, the authors evaluated the performance of their motion signal analysis algorithm, which was then compared with the state-of-the-art approach based on MoSIFT descriptors with Bag-of-Words mechanism, and basic motion signal analysis with Bag-of-Words.

The experimental results indicate that the proposed approach accurately detects fights in real scenarios and performs better than the MoSIFT approach.

By collecting and annotating real surveillance videos containing real fight events and augmenting with well-known data sets, the authors proposed, implemented and evaluated a low computation approach, comparing it with the state-of-the-art approach. The authors uncovered some fundamental differences between real and simulated fights and initiated a new study in discriminating real against simulated fight events, with very good performance.

Background: Geostatistics focuses on spatial or spatiotemporal datasets. Geostatistics was initially developed to generate probability distribution predictions of ore grade in the mining industry; however, it has been successfully applied in diverse scientific disciplines. This technique includes univariate, multivariate, and simulations. Kriging geostatistical methods, simple, ordinary, and universal Kriging, are not multivariate models in the usual statistical function. Notwithstanding, simple, ordinary, and universal kriging techniques utilize random function models that include unlimited random variables while modeling one attribute. The coKriging technique is a multivariate estimation method that simultaneously models two or more attributes defined with the same domains as coregionalization.

Objective: This study investigates the impact of populations on traffic volumes as a variable. The additional variable determines the strength or accuracy obtained when data integration is adopted. In addition, this is to help improve the estimation of annual average daily traffic (AADT).

Methods procedures, process: The investigation adopts the coKriging technique with AADT data from 2009 to 2016 from Montana, Minnesota, and Washington as primary attributes and population as a controlling factor (second variable). CK is implemented for this study after reviewing the literature and work completed by comparing it with other geostatistical methods.

Results, observations, and conclusions: The Investigation employed two variables. The data integration methods employed in CK yield more reliable models because their strength is drawn from multiple variables. The cross-validation results of the model types explored with the CK technique successfully evaluate the interpolation technique's performance and help select optimal models for each state. The results from Montana and Minnesota models accurately represent the states' traffic and population density. The Washington model had a few exceptions. However, the secondary attribute helped yield an accurate interpretation. Consequently, the impact of tourism, shopping, recreation centers, and possible transiting patterns throughout the state is worth exploring.

The purpose of this paper is to address a solution to the problem of defect recognition from images using the support vector machines (SVM).

A SVM‐based multi‐class pattern recognition system has been developed for inspecting commonly occurring fabric defects such as neps, broken ends, broken picks and oil stain. A one‐leave‐out cross validation technique is applied to assess the accuracy of the SVM classifier in classifying fabric defects.

The investigation indicates that the fabric defects can be classified with a reasonably high degree of accuracy by the proposed method.

The paper outlines the theory and application of SVM classifier with reference to pattern classification problem in textiles. The SVM classifier outperforms the other techniques of machine learning systems such as artificial neural network in terms of efficiency of calculation. Therefore, SVM classifier has great potential for automatic inspection of fabric defects in industry.

The issue of model selection in applied research is of vital importance. Since the true model in such research is not known, which model should be used from among various potential ones is an empirical question. There might exist several competitive models. A typical approach to dealing with this is classic hypothesis testing using an arbitrarily chosen significance level based on the underlying assumption that a true null hypothesis exists. In this paper, the authors investigate how successful the traditional hypothesis testing approach is in determining the correct model for different data generating processes using time series data. An alternative approach based on more formal model selection techniques using an information criterion or cross-validation is also investigated.

Monte Carlo simulation experiments on various generating processes are used to look at the response surfaces resulting from hypothesis testing and response surfaces resulting from model selection based on minimizing an information criterion or the leave-one-out cross-validation prediction error.

The authors find that the minimization of an information criterion can work well for model selection in a time series environment, often performing better than hypothesis-testing strategies. In such an environment, the use of an information criterion can help reduce the number of models for consideration, but the authors recommend the use of other methods also, including hypothesis testing, to determine the appropriateness of a model.

This paper provides an alternative approach for selecting the best potential model among many for time series data. It demonstrates how minimizing an information criterion can be useful for model selection in a time-series environment in comparison to some standard hypothesis testing strategies.

Pressure ulcer is a clinical pathology of localized damage to the skin and underlying tissue caused by pressure, shear, and friction. Diagnosis, treatment and care of pressure ulcers involve high costs for sanitary systems. Accurate wound evaluation is a critical task to optimize the efficacy of treatments and health‐care. Clinicians evaluate the pressure ulcers by visual inspection of the damaged tissues, which is an imprecise manner of assessing the wound state. Current computer vision approaches do not offer a global solution to this particular problem. The purpose of this paper is to use a hybrid learning approach based on neural and Bayesian networks to design a computational system to automatic tissue identification in wound images.

A mean shift procedure and a region‐growing strategy are implemented for effective region segmentation. Color and texture features are extracted from these segmented regions. A set of k multi‐layer perceptrons is trained with inputs consisting of color and texture patterns, and outputs consisting of categorical tissue classes determined by clinical experts. This training procedure is driven by a k‐fold cross‐validation method. Finally, a Bayesian committee machine is formed by training a Bayesian network to combine the classifications of the k neural networks (NNs).

The authors outcomes show high efficiency rates from a two‐stage cascade approach to tissue identification. Giving a non‐homogeneous distribution of pattern classes, this hybrid approach has shown an additional advantage of increasing the classification efficiency when classifying patterns with relative low frequencies.

The methodology and results presented in this paper could have important implications to the field of clinical pressure ulcer evaluation and diagnosis.

The novelty associated with this work is the use of a hybrid approach consisting of NNs and Bayesian classifiers which are combined to increase the performance of a pattern recognition task applied to the real clinical problem of tissue detection under non‐controlled illumination conditions.

To investigate the feasibility of using artificial neural networks for conceptual design of bracings systems for tall steel buildings.

Database of 234 design examples has been developed using commercially available detailed design software. These examples represent building up to 20 storeys. Feed forward back‐propagation neural network is trained on these examples. The results obtained from the artificial neural network are evaluated by re‐substitution, hold‐out and ten‐fold cross‐validation techniques.

Results indicate that artificial neural network would give a performance of 97.91 percent (ten‐fold cross‐validation). The performance of this system is benchmarked by developing a binary logistic regression model from the same data. Performance of the two models has been compared using McNemar's test and receiver operation characteristics curves. Artificial neural network shows a better performance. The difference is found to be statically significant.

The developed model is applicable only to steel building up to 20 storeys. The feasibility of using artificial neural networks for conceptual design of bracings systems for tall steel buildings more than 20 storeys has not been investigated.

Implementation of the broad methodology outlined for the use of neural networks can be accomplished by conducting short training courses. This will provide personnel with flexibility in addressing buildings‐specifics bracing conditions and limitations.

In tall building design a lot of progress has been made in the development of software tools for numerical intensive tasks of analysis, design and optimization, however, professional software tools are not available to help the designer to choose an optimum building configuration at the conceptual design stage. The presented research provides a methodology to investigate the feasibility of using artificial neural networks for conceptual design of bracings systems for tall buildings. It is found that this approach for the selection of bracings in tall buildings is a better and cost effective option compared with database generated on the basis of expert opinion. It also correctly classifies and recommends the type of trussed bracing system.

Successful bridge management system (BMS) development requires a reliable bridge deterioration model, which is the most crucial component in a BMS. Historical condition ratings obtained from biennial bridge inspections are a major source for predicting future bridge deterioration in BMSs. However, historical condition ratings are very limited in most bridge agencies, thus posing a major barrier for predicting reliable future bridge performance. The purpose of this paper is to present a preliminary study as part of a long‐term research on the development of a reliable bridge deterioration model using advanced Artificial Intelligence (AI) techniques.

This proposed study aims to develop a reliable deterioration model. The development work consists of two major Stages: stage 1 – generating unavailable bridge element condition rating records using the Backward Prediction Model (BPM). This helps to provide sufficient historical deterioration patterns for each element; and stage 2 – predicting long‐term condition ratings based on the outcome of Stage 1 using time delay neural networks (TDNNs).

Long‐term prediction using proposed method can also be expressed in the same form of inspection records – element quantities of each bridge element can be predicted. The proposed AI‐based deterioration model does not ignore critical failure risks in small number of bridge elements in low condition states (CSs). This implies that the risk in long‐term predictions can be reduced.

The proposed methodology aims to utilise limited bridge inspection records over a short period to predict large datasets spanning over a much longer time period for a reliable, accurate and efficient long‐term bridge deterioration model. Typical uncertainty, due to the limitation of overall condition rating (OCR) method, can be minimised in long‐term predictions using limited inspection records.