Search

Label noise poses a significant challenge to supervised learning algorithms. Extensive research has been conducted on classification and regression tasks, but label noise filtering methods specifically designed for ordinal regression are lacking. In this paper, we propose a set of ordinal label noise filtering frameworks by theoretically exploring the generalization error bound in noisy environments. Besides, we present a robust label noise estimation method voted by inter-class distance. It takes into account the nonuniformity of ordinal labels and the reliability of the base model. This estimator is integrated into our framework in the proposed Inter-Class Distance-based Filtering (ICDF) algorithm. We empirically demonstrate the effectiveness of ICDF in identifying label noise and achieving improved generalization performance. Our experiments conducted on benchmark and real age estimation datasets show the superiority of ICDF over the existing filters in ordinal label noise cleaning. [ABSTRACT FROM AUTHOR]

Semi-supervised ordinal regression (S2OR) has been recognized as a valuable technique to improve the performance of the ordinal regression (OR) model by leveraging available unlabeled samples. The balancing constraint is a useful approach for semi-supervised algorithms, as it can prevent the trivial solution of classifying a large number of unlabeled examples into a few classes. However, rapid training of the S2OR model with balancing constraints is still an open problem due to the difficulty in formulating and solving the corresponding optimization objective. To tackle this issue, we propose a novel form of balancing constraints and extend the traditional convex–concave procedure (CCCP) approach to solve our objective function. Additionally, we transform the convex inner loop (CIL) problem generated by the CCCP approach into a quadratic problem that resembles support vector machine, where multiple equality constraints are treated as virtual samples. As a result, we can utilize the existing fast solver to efficiently solve the CIL problem. Experimental results conducted on several benchmark and real-world datasets not only validate the effectiveness of our proposed algorithm but also demonstrate its superior performance compared to other supervised and semi-supervised algorithms [ABSTRACT FROM AUTHOR]

Emotional understanding and expression plays a critical role in social interaction. To analyze children's emotional interaction automatically, this study focuses on developing a novel network architecture and a reliable algorithm for expression intensity estimation to measure children's facial expression responses to emotional stimuli. The facial expression intensity variation provides temporal dynamic information of facial behavior, which is critical to interpreting the meaning of expression. In order to avoid laborious manual annotations for expression intensity, existing unsupervised methods attempt to identify relative intensity using ordinal information within a facial expression sequence; however, they fail to estimate absolute intensity accurately. Moreover, appropriate features are needed to represent the continuous appearance changes caused by expression intensity to improve the model's ability to distinguish subtle differences in expression. This study therefore presents a novel semi-supervised method to estimate expression intensity using salient deep learning features. First, the facial expression is represented by the difference response of the convolutional neural network backbone between the target expression and its responding neutral expression, with the goal of suppressing the effects of expression-unrelated features on expression intensity estimation. Then, the pairwise data constructed with ordinal information is input into a Siamese network with a combined hinge loss that guides learning the relative intensity on unlabeled pairwise frames, the absolute intensity of a few labeled key frames, and the intensity range of most unlabeled frames. The average pearson correlation coefficient, intraclass correlation coefficient, and mean absolute error are 0.7683, 0.7405, and 0.1698 on the extended Cohn-Kanade dataset (CK+), and 0.7804, 0.6684, and 0.1864 on the Binghamton University 4D Facial Expression Dataset using the proposed method, results that are superior to the state of the art. The cross-dataset experiment indicates that the proposed method is promising for the analysis of children's emotional interactions. [ABSTRACT FROM AUTHOR]

Facial expression intensity estimation has promising applications in health care and affective computing, such as monitoring patients’ pain feelings. However, labeling facial expression intensity is a specialized and time-consuming task. Ordinal regression (OR)-based methods address this issue to some extent by estimating the relative intensity but failing to estimate the absolute intensity due to lack of exploring useful information from noisy labels caused by manual and automatic labeling biases. Inspired by label distribution learning (LDL) to resist the noisy labels, this paper introduces the label-distribution-learning-enhanced OR (LDL-EOR) approach for facial expression intensity estimation. This design aims to utilize LDL to improve the accuracy of absolute intensity estimation while keeping the cost of manual labeling low. The label distribution is converted into a continuous intensity value by calculating the mathematical expectation, which makes the prediction results meet both relative and absolute intensity constraints. Ensuring the feasibility of LDL-EOR in different supervised settings, this paper presents a unified label distribution generation framework to automatically relabel training data frame by frame. The generated soft labels are used to supervise the LDL-EOR model and enhance its robustness to the noise existing in the original labels. Numerous experiments were conducted on three public expression datasets (CK+, BU-4DFE, and PAIN) to validate the superiority of LDL-EOR relative to other state-of-the-art approaches. [ABSTRACT FROM AUTHOR]

Motivated by data measuring progression of leishmaniosis in a cohort of US dogs, we develop a Bayesian longitudinal model with autoregressive errors to jointly analyze ordinal and continuous outcomes. Multivariate methods can borrow strength across responses and may produce improved longitudinal forecasts of disease progression over univariate methods. We explore the performance of our proposed model under simulation, and demonstrate that it has improved prediction accuracy over traditional Bayesian hierarchical models. We further identify an appropriate model selection criterion. We show that our method holds promise for use in the clinical setting, particularly when ordinal outcomes are measured alongside other variables types that may aid clinical decision making. This approach is particularly applicable when multiple, imperfect measures of disease progression are available. [ABSTRACT FROM AUTHOR]

Accurate maturity identification of permanent teeth for children and adolescents using orthopantomogram (OPG) images is crucial in pediatric dentistry.Although recent advancements in deep learning have shown promising outcomes in maturity staging of third molars and mandibular permanent dentition, limited research has focused on assessing the entire dentition. Furthermore, accurate performance is hindered due to variations in tooth appearance and similarities between adjacent stages. This paper develops an automatic methodology to identify the tooth maturity stages of the full permanent dentition. Specifically, a tooth localization model based on YOLOv3 is presented to detect and number each candidate permanent tooth. Subsequently, we propose a novel symmetric and ordinal staging network (SOS-Net) for the precise maturity identification of each tooth. SOS-Net incorporates a symmetric feature combination block (SFC Block) based on bilateral facial symmetry to enhance discriminative feature representation in order to improve performance, while minimizing increases in model parameters. Additionally, we introduce an auxiliary regression branch with a novel loss called adjacent stage aware (ASA) loss to provide information in the order of maturity levels and effectively reduce misclassification between adjacent stages. The proposed methodology is evaluated on a private OPG image dataset collected from individuals aged 3 to 14 years. The ablation studies and visual analysis of SOS-Net verified the effectiveness and interpretability of the SFC block and ASA loss, with a F1-score improvement of 3.77 % . Moreover, comparative experiments demonstrate the good performance of the proposed methodology compared to state-of-the-art methods in maturity staging and age estimation. [ABSTRACT FROM AUTHOR]

Diabetic retinopathy (DR) is a chronic complication of diabetes that damages the retinal blood vessels, leading to impaired vision and even blindness, and is one of the top three eye diseases causing human blindness. An effective DR grading algorithm can help ophthalmologists to diagnose patients and improve efficiency. Therefore, we propose a fusion network based on transformer and convolutional neural network (CNN) to perform DR grading. The proposed approach addresses two critical issues in the task: i) The CNN-based DR classification method has a small receptive field, so the range of available information is limited. The Transformer-based DR classification method has a large receptive field, but it is easy to lose local details; ii) Existing approaches treat DR grading as a traditional multiclass classification task, which ignores ordinal information between DR of different levels. To address i), we fuse the multi-level features of CNN and Transformer at different stages, and realize the enhancement and interaction of local and global information through the fusion module. To address ii), we propose a KC loss by formulating DR grading as a joint ordinal regression and multiclass classification problem, and obtain both category-supervised information and ordinal supervised information. Extensive qualitative and quantitative assessments have shown that our approach achieves superior performance on publicly available DeepDR and IDRiD datasets. [ABSTRACT FROM AUTHOR]

In recent times, deep neural networks achieved outstanding predictive performance on various classification and pattern recognition tasks. However, many real-world prediction problems have ordinal response variables, and this ordering information is ignored by conventional classification losses such as the multi-category cross-entropy. Ordinal regression methods for deep neural networks address this. One such method is the CORAL method, which is based on an earlier binary label extension framework and achieves rank consistency among its output layer tasks by imposing a weight-sharing constraint. However, while earlier experiments showed that CORAL's rank consistency is beneficial for performance, it is limited by a weight-sharing constraint in a neural network's fully connected output layer, which may restrict the expressiveness and capacity of a network trained using CORAL. We propose a new method for rank-consistent ordinal regression without this limitation. Our rank-consistent ordinal regression framework (CORN) achieves rank consistency by a novel training scheme. This training scheme uses conditional training sets to obtain the unconditional rank probabilities through applying the chain rule for conditional probability distributions. Experiments on various datasets demonstrate the efficacy of the proposed method to utilize the ordinal target information, and the absence of the weight-sharing restriction improves the performance substantially compared to the CORAL reference approach. Additionally, the suggested CORN method is not tied to any specific architecture and can be utilized with any deep neural network classifier to train it for ordinal regression tasks. [ABSTRACT FROM AUTHOR]

Label distribution learning achieved promising results on ordinal regression tasks such as facial age and attractiveness estimation, especially using deep label distribution learning (DLDL) methods, introducing the label distribution learning into deep convolutional neural networks. However, existing DLDL methods have an inconsistency between the training objectives and the evaluation metric, so they may be suboptimal. In addition, these methods always adopt image classification or face recognition models with a large amount of parameters, which carry expensive computation cost and storage overhead. In this paper, we firstly analyze the essential relationship between two state-of-the-art methods (ranking CNN and DLDL) and show that the ranking method is in fact learning label distribution implicitly. This result thus firstly unifies two existing popular state-of-the-art methods into the DLDL framework. Second, in order to alleviate the inconsistency and reduce resource consumption, we design a lightweight network architecture and propose a unified framework which can jointly learn label distribution and regress expectation value. The effectiveness of our approach has been demonstrated on typical ordinal regression tasks including facial age and attractiveness estimation. Our method achieves new state-of-the-art results using the single model with 36× fewer parameters and 3× faster inference speed. [ABSTRACT FROM AUTHOR]

The assessment of bone age is important for evaluating child development, optimizing the treatment for endocrine diseases, etc. And the well-known Tanner-Whitehouse (TW) clinical method improves the quantitative description of skeletal development based on setting up a series of distinguishable stages for each bone individually. However, the assessment is affected by rater variability, which makes the assessment result not reliable enough in clinical practice. The main goal of this work is to achieve a reliable and accurate skeletal maturity determination by proposing an automated bone age assessment method called PEARLS, which is based on the TW3-RUS system (analysis of the radius, ulna, phalanges, and metacarpal bones). The proposed method comprises the point estimation of anchor (PEA) module for accurately localizing specific bones, the ranking learning (RL) module for producing a continuous stage representation of each bone by encoding the ordinal relationship between stage labels into the learning process, and the scoring (S) module for outputting the bone age directly based on two standard transform curves. The development of each module in PEARLS is based on different datasets. Finally, corresponding results are presented to evaluate the system performance in localizing specific bones, determining the skeletal maturity stage, and assessing the bone age. The mean average precision of point estimation is 86.29%, the average stage determination precision is 97.33% overall bones, and the average bone age assessment accuracy is 96.8% within 1 year for the female and male cohorts. [ABSTRACT FROM AUTHOR]

Purpose: Socioeconomic factors are known to modulate health. Concerning sleep apnea, influences of income, education, work, and living in a partnership are established. However, results differ between national and ethnic groups. Results also differ between various clinical studies and population-based approaches. The goal of our study was to determine if such factors can be verified in the population of Pomerania, Germany. Methods: A subgroup from the participants of the population-based Study of Health in Pomerania volunteered for an overnight polysomnography. Their data were subjected to an ordinal regressions analysis with age, sex, body mass index (BMI), income, education, work, and life partner as predictors for the apnea–hypopnea index. Results: Among the subgroup (N = 1209) from the population-based study (N = 4420), significant effects were found for age, sex, and BMI. There were no significant effects for any of the socioeconomic factors. Conclusion: Significant effects for well-established factors as age, sex, and BMI show that our study design has sufficient power to verify meaningful associations with sleep apnea. The lack of significant effects for the socioeconomic factors suggests their clinical irrelevance in the tested population. [ABSTRACT FROM AUTHOR]

Public expectations of forests as high-quality restorative environments that facilitate subjective well-being and stress relief along with numerous health benefits have been rising sharply during recent decades. In addition, the COVID-19 pandemic and its accompanying restrictive measures also transformed forests into some of the few places to spend time away from home. The presented study drew on the assumption that the pandemic situation and a rise in the number of forest visits would affect the experience, recognition, and appreciation of the well-being aspects related to spending time in forests. The study goal was to elucidate the potential effects of the COVID-19 pandemic on the relationships between forest visits, well-being and stress relief, emotions, perception of nature and forest value and importance, pro-environmental behavior, and societal expectations of the role of forests and forest ecosystem services. A survey using a digital questionnaire was conducted several months after the pandemic outbreak on a representative sample of the Slovak population. The Wilcoxon test and ordinal regression analysis were used to identify significant relationships, e.g., between the recency of anger episodes and the number of forest visits. The results showed that the pandemic strengthened the perception of forests as a high-quality restorative environment and that emotions associated with forest visits played an important role in the perceived importance of forests and their possible overexploitation. The results underscore the urgent need to put demands for forest recreation on par with the forest bioeconomy and to sensitize forest visitors to management and conservation requirements. [ABSTRACT FROM AUTHOR]

Face anti-spoofing is a crucial link to ensure the security of face recognition. This paper proposes a novel face anti-spoofing method, which performs ordinal regression with representative feature strengthening to learn generalized and discriminative representation for the live and spoof face classification. Specifically, we propose a semantic label schema, which encodes the inter-class ordinal relationships among live and various spoof faces into supervision information to supervise deep neural networks to perform ordinal regression. It enables the learned model to finely constrain the relative distances among features of different categories in the feature space according to the ordinal relationships. The representative feature strengthening network is designed to strengthen important features and meanwhile weaken redundant features for the classification decision. It leverages a dual-task architecture that takes the same single image as input and shares representations via feature fusing blocks. The network first fuses hierarchical paired convolutional features of two streams to learn the common concern of the two related tasks and then, aggregates the learned local convolutional features into a global representation by a learnable feature weighting block. The network is trained to minimize the Kullback–Leibler divergence loss in an end-to-end manner supervised by the semantic labels. Extensive intra-dataset and cross-dataset experiments demonstrate that the proposed method outperforms the state-of-the-art approaches on four widely used face anti-spoofing datasets. [ABSTRACT FROM AUTHOR]

Existing ordinal trees and random forests typically use scores that are assigned to the ordered categories, which implies that a higher scale level is used. Versions of ordinal trees are proposed that take the scale level seriously and avoid the assignment of artificial scores. The construction principle is based on an investigation of the binary models that are implicitly used in parametric ordinal regression. These building blocks can be fitted by trees and combined in a similar way as in parametric models. The obtained trees use the ordinal scale level only. Since binary trees and random forests are constituent elements of the proposed trees, one can exploit the wide range of binary trees that have already been developed. A further topic is the potentially poor performance of random forests, which seems to have been neglected in the literature. Ensembles that include parametric models are proposed to obtain prediction methods that tend to perform well in a wide range of settings. The performance of the methods is evaluated empirically by using several data sets. [ABSTRACT FROM AUTHOR]

This work focuses on learning causal network structures from ordinal categorical data. By combining constraint-based with score-and-search methodologies in structural learning, we propose a hybrid method called Markov Blanket Based Ordinal Causal Discovery (MBOCD) algorithm, which can capture the ordinal relationship of values in ordinal categorical variables. Theoretically, it is proved that for ordinal causal networks, two adjacent DAGs belonging to the same Markov equivalence class are identifiable, which results in the generation of a causal graph. Simulation experiments demonstrate that the proposed algorithm outperforms existing methods in terms of computational efficiency and accuracy. The code of this work is open at: https://github.com/leoydu/MBOCDcode.git. [ABSTRACT FROM AUTHOR]

The purpose of this paper is to introduce a new distance metric learning algorithm for ordinal regression. Ordinal regression addresses the problem of predicting classes for which there is a natural ordering, but the real distances between classes are unknown. Since ordinal regression walks a fine line between standard regression and classification, it is a common pitfall to either apply a regression-like numerical treatment of variables or underrate the ordinal information applying nominal classification techniques. On a different note, distance metric learning is a discipline that has proven to be very useful when improving distance-based algorithms such as the nearest neighbors classifier. In addition, an appropriate distance can enhance the explainability of this model. In our study we propose an ordinal approach to learning a distance, called chain maximizing ordinal metric learning. It is based on the maximization of ordered sequences in local neighborhoods of the data. This approach takes into account all the ordinal information in the data without making use of any of the two extremes of classification or regression, and it is able to adapt to data for which the class separations are not clear. We also show how to extend the algorithm to learn in a non-linear setup using kernel functions. We have tested our algorithm on several ordinal regression problems, showing a high performance under the usual evaluation metrics in this domain. Results are verified through Bayesian non-parametric testing. Finally, we explore the capabilities of our algorithm in terms of explainability using the case-based reasoning approach. We show these capabilities empirically on two different datasets, experiencing significant improvements over the case-based reasoning with the traditional Euclidean nearest neighbors. [ABSTRACT FROM AUTHOR]

Many real-world datasets are labeled with natural orders, i.e., ordinal labels. Ordinal regression is a method to predict ordinal labels that finds a wide range of applications in data-rich domains, such as natural, health and social sciences. Most existing ordinal regression approaches work well for independent and identically distributed (IID) instances via formulating a single ordinal regression task. However, for heterogeneous non-IID instances with well-defined local geometric structures, e.g., subpopulation groups, multi-task learning (MTL) provides a promising framework to encode task (subgroup) relatedness, bridge data from all tasks, and simultaneously learn multiple related tasks in efforts to improve generalization performance. Even though MTL methods have been extensively studied, there is barely existing work investigating MTL for heterogeneous data with ordinal labels. We tackle this important problem via sparse and deep multi-task approaches. Specifically, we develop a regularized multi-task ordinal regression (MTOR) model for smaller datasets and a deep neural networks based MTOR model for large-scale datasets. We evaluate the performance using three real-world healthcare datasets with applications to multi-stage disease progression diagnosis. Our experiments indicate that the proposed MTOR models markedly improve the prediction performance comparing with single-task ordinal regression models. [ABSTRACT FROM AUTHOR]

In ornamental aquaculture, the traits related with beauty are an important factor in the economy and their inclusion in the genetic improvement program is essential. However, in some species, it is difficult to define which traits are important in this sense. The objective of this study was to know the producers', traders', and aquarists' interests of morphometric characteristics associated with the esthetic of Puntius tetrazona for a genetic improvement program. The information was generated from surveys considering aquarists, traders, and producers. The design of the questionnaires consisted of sets of questions according to the type of information that we wanted to obtain: demographic, specific to each actor, and esthetic preferences; the last was a common section for the three actors. The last one is supported by five sets of morphotypes based on the morphometric characteristics with a major contribution to variability in Puntius tetrazona. For the analysis of the preferences of each population for every group of morphotypes, an accumulated ordinal model was utilized. The perception of beauty is not affected by demographic traits. And the morphotypes with a higher preference for inclusion in a selection program to esthetic traits in P. tetrazona would be a short caudal peduncle, medium-length caudal fin, anal fin that is short proximal long and long distal long, wide body marks, and dorsal fin that is 25% gray. For the implementation of genetic improvement programs, it will be important to take into account the differences in interests at two points within the production chain, the needs of the producer and those of the aquarists. [ABSTRACT FROM AUTHOR]

Longitudinal studies often involve multiple mixed response variables measured repeatedly over time. Although separate modeling of these multiple mixed response variables can be easily performed, they may lead to inefficient estimates and consequently, misleading inferences. For obtaining correct inference, one needs to model multiple mixed responses jointly. In this paper, we use copula models for defining a multivariate distribution for multiple mixed outcomes at each time point. Then, we use transition model for considering association between longitudinal measurements. Two simulation studies are performed for illustration of the proposed approach. The results of the simulation studies show that the use of the separate models instead of the joint modeling leads to inefficient parameter estimates. The proposed approach is also used for analyzing two real data sets. The first data set is a part of the British Household Panel Survey. In this data set, annual income and life satisfaction are considered as the continuous and the ordinal correlated longitudinal responses, respectively. The second data set is a longitudinal data about heart failure patients. This study is a treatment–control study, where the effect of a treatment is simultaneously investigated on readmission and referral to doctor as two binary associated longitudinal responses. [ABSTRACT FROM AUTHOR]

Ordinal regression (OR) or classification is a machine learning paradigm for ordinal labels. To date, there have been a variety of methods proposed including kernel based and neural network based methods with significant performance. However, existing OR methods rarely consider latent structures of given data, particularly the interaction among covariates, thus losing interpretability to some extent. To compensate this, in this paper, we present a new OR method: ordinal factorization machine with hierarchical sparsity (OFMHS), which combines factorization machine and hierarchical sparsity together to explore the hierarchical structure behind the input variables. For the sake of optimization, we formulate OFMHS as a convex optimization problem and solve it by adopting the efficient alternating directions method of multipliers (ADMM) algorithm. Experimental results on synthetic and real datasets demonstrate the superiority of our method in both performance and significant variable selection. [ABSTRACT FROM AUTHOR]

Herbarium data have availed the much-needed long-term data that have constrained long-term and real-time phenological studies. Using herbarium specimens from the National Herbarium of Namibia, species, region, rainfall, latitude and longitude were regressed against the day of flowering to determine the effect of each variable on the day of flowering. This study used parameter estimates and the odds ratios from ordinal regression analysis to interpret the influence of the explanatory variables on the day of flowering. The results revealed that flowering dates of C. lugardiae have advanced by six days/decade during the period 1950 to 2010. Average summer rainfall decreased by approximately 3 mm per year from 1960 to 2015 in Namibia. Consequently, significant rainfall effects were recorded on the day of flowering for C. buphanoides, C. stuhlmannii subsp. delagoense and C. paludosum which have advanced by approximately 1 day/decade/mm, 1 day/decade/mm and 2 day/decade/mm during the period studied, respectively. Increase in latitude delayed day of flowering whilst an advancement of the day of flowering with an increase in longitude was observed. Substantial interaction effects among the explanatory variables were demonstrated. Combined effects of rainfall and latitude had a negative effect on the day of flowering whereas interaction effects of average summer rainfall and longitude had a positive combined effect on day of flowering. Our results, therefore, reveal the complex effects of rainfall and geographical location on flowering phenology of different Crinum species. The potential impacts of the phenological responses of different Crinum species on other organisms are discussed. [ABSTRACT FROM AUTHOR]

Global university rankings have become a critical factor in the higher education sector, engendering increasing interest and exerting a notable influence over a wide variety of stakeholders. They are presented to different audiences as tools that evaluate and rank universities according to quality. However, some authors are of the opinion that rankings express reputational factors to a large extent. This article presents a model of the intra- and inter-ranking relationships from the perspective of reputation along with an empirical study on two of the most influential rankings: the Academic Ranking of World Universities and the Times Higher Education World University Rankings. Data from these two rankings between 2010 and 2018, and the application of ordinal regressions, provide evidence that both rankings are mutually influential, generating intra and reciprocal reputational effects over time. [ABSTRACT FROM AUTHOR]

Imbalanced classification has been extensively researched in the last years due to its prevalence in real-world datasets, ranging from very different topics such as health care or fraud detection. This literature has long been dominated by variations of the same family of solutions (e.g. mainly resampling and cost-sensitive learning). Recently, a new and promising way of tackling this problem has been introduced: learning with scoring pairwise ranking so that each pair of classes contribute in tandem to the decision boundary. In this sense, the paper addresses the problem of class imbalance in the context of ordinal regression, proposing two novel contributions: (a) approaching the imbalance by binary pairwise ranking using a well-known label decomposition ensemble, and (b) introducing a regularization into this ensemble so that parallel decision boundaries are favored. These are two independent contributions that synergize well. Our model is tested using linear Support Vector Machines and our results are compared against state-of-the-art models. Both approaches show promising performance in ordinal class imbalance, with an overall 15% improvement relative to the state-of-the-art, as evaluated by a balanced metric. [ABSTRACT FROM AUTHOR]

On the basis of the Impact Factor of Journal Citation Reports developed by ISI as a journal quality indicator, this paper puts forth an ordinal regression model to estimate the journal’s position by terciles. The set of explanatory variables includes the H-index of its Editor-in-chief, percentage of papers published in the journal that received external funding, average number of papers published yearly, and two factors concerning the scope and structure of the journal. The proposed model was applied to the field of Dentistry, Oral Surgery and Medicine, and led us to the conclusion that the above mentioned covariables alone had a significant input in the model, but not the factors. The essay performed on a sample of 30 Dentistry journals included in JCR provided a confirmatory correct classification rate (CCR) of 80%, with a predictive CCR of 75% on a sample of eight new journals not previously considered in the phase of model estimation. [ABSTRACT FROM AUTHOR]

In some applications of machine learning and information retrieval (e.g. medical diagnosis, image recognition, pre-classification...), it can be preferable to provide less informative but more reliable predictions. This can be done by making partial predictions in the form of class subsets when the available information is insufficient to provide a reliable unique class. Imprecise probabilistic approaches offer nice tools to learn models from which such cautious predictions can be produced. However, the learning and inference processes of such models are computationally harder than their precise counterparts. In this paper, we introduce and study a particular binary decomposition strategy, nested dichotomies, that offer computational advantages in both the learning (due to the binarization process) and the inference (due to the decomposition strategy) processes. We show with experiments that these computational advantages do not lower the performances of the classifiers, and can even improve them when the class space has some structure. [ABSTRACT FROM AUTHOR]

Contravening findings are found in the literature about the association between individuals stature and diabetes status. This paper attempts to find evidence regarding the nature and strength of this association by analyzing a nationally representative data of Bangladesh. Under ordinal regression setup, a partial proportional odds model, a variant of the proportional odds model, has been found to fit better compared to the proportional odds model for analyzing diabetes status in Bangladesh. A strong evidence of inverse relationship is found between diabetes and height. Gender, education level, wealth index, place of residence, division and type of work are also found to be significantly associated with diabetes. [ABSTRACT FROM AUTHOR]

The support vector ordinal regression (SVOR) method is derived from support vector machine and developed to tackle the ordinal regression problems. However, it ignores the distribution characteristics of the data. In this paper, we propose a novel method to handle the ordinal regression problems. This method is referred to as minimum class variance support vector ordinal regression (MCVSVOR). In contrast with SVOR, MCVSVOR explicitly takes into account the distribution of the categories and achieves better generalization performance. Moreover, the problem of MCVSVOR can be transformed into one of SVOR. Thus, the existing software of SVOR can be used to solve the problem of MCVSVOR. In the paper, we first discuss the linear case of MCVSVOR and then develop the nonlinear MCVSVOR through using the kernelization trick. The comprehensive experiment results show that the proposed method is effective and can achieve better generalization performance in contrast with SVOR. [ABSTRACT FROM AUTHOR]

When having to make a prediction under probabilistic uncertainty in ordinal problems, the median offers a number of interesting properties compared to other statistics such as the expected value. In particular, it does not depend on a particular metric defined over the elements, but still takes account of the ordinal nature of the data. It can also be shown to be the minimizer of the $$L_1$$ loss function. In this paper, we show that similar results can be obtained when the uncertainty is described not by a single probability distribution, but by a convex set of those. In particular, we relate the lower and upper medians to the $$L_1$$ loss function via the notion of lower and upper expectations (and extend these results to general quantiles). We also show that, using a different decision rule, the lower and upper median can be retrieved when assuming the cost to be strictly monotonic and symmetric, and nothing more. Finally, we run some tests to show the interest of using Median based predictions with convex sets of probabilities in ordinal regression problems. [ABSTRACT FROM AUTHOR]

Ordinal regression is a kind of regression analysis used for predicting an ordered response variable. In these problems, the patterns are labelled by a set of ranks with an ordering among the different categories. The most common type of ordinal regression model is the cumulative link model. The cumulative link model relates an unobserved continuous latent variable with a monotone link function. Logit and probit functions are examples of link functions used in cumulative link models. In this paper, a novel generalized link function based on a generalization of the logistic distribution is proposed. The generalized link function proposed is able to reproduce other different link functions by changing two real parameters: $$\alpha $$ and $$\lambda $$ . The generalized link function has been included in a cumulative link model where the latent function is determined by a standard neural network in order to test the performance of the proposal. For this model, a reformulation of the tunable thresholds and distribution parameters was applied to convert the constrained optimization problem into an unconstrained optimization problem. Experimental results demonstrate that our proposed approach can achieve competitive generalization performance. [ABSTRACT FROM AUTHOR]

The pattern of increased nest defense effort over the course of a nesting season could result from three distinct (albeit non-exclusive) mechanisms: increased value of offspring to parents with progression toward independence (parental investment theory), decreased opportunity for renesting (renesting potential hypothesis), or decreased perceived costs of defense after repeated encounters with human observers (positive reinforcement hypothesis). To gauge relative empirical support for each of these mechanisms, we disentangle these three often-confounded hypotheses using multimodel inference with mixed-model ordinal regression applied to an extensive red-winged blackbird ( Agelaius phoeniceus) nesting data set (4518 monitoring visits to 1330 nests). Parent aggression was rated on an ordinal scale (0-4) during repeated monitoring visits. Additionally, we assessed clutch/brood size, nest density, time of day, and nest concealment effects on aggression. In a preliminary analysis, including all three major hypotheses, male and female nest defense was most strongly explained by parental investment (nest age). Positive reinforcement (visit number) and renesting potential (Julian date) were also well-supported predictors in males. The interactions of decomposed nest age (within-individual and between-individual centered) with Julian date were particularly important in the top male model. Additional factors, such as clutch/brood size, nest density, and nest concealment, appeared to have larger predictive roles in explaining female aggression relative to males. These patterns are likely explained by different sexual reproductive roles within a polygynous mating system. Our study highlights the importance of interacting mechanisms involving parental investment theory and the use of within-individual standardization to help disentangle competing and empirically confounded hypotheses. Significance statement: Avian nest defense generally increases over the course of a nesting season, potentially from the result of three different mechanisms: parental investment theory, renesting potential hypothesis, or positive reinforcement hypothesis from repeated nest visitation. We revisit this classic question through a comprehensive analytical approach with an extensive observational data set with red-winged blackbirds, employing multimodel selection and within-individual and between-individual centering techniques. We found that parental investment (nest age) was the strongest predictor of nest defense for both sexes; however, positive reinforcement and renesting potential also appeared to help explain additional variation in nest defense for males. Competitiveness of models with interactive effects indicated that these mechanisms do not operate independently for either sex, and additional covariates (e.g., clutch/brood size) especially aided female model competiveness. Our study highlights the importance of multiple and often interacting factors that influence avian nest defense. [ABSTRACT FROM AUTHOR]

In this paper, an ordinal regression model and a K-means clustering method are employed to directly assess the intensity of snow, low-temperature and frost disasters in non-pasture areas. Then, economic loss caused by the disasters is evaluated quantitatively by static and dynamic input-output models, respectively. Finally, an empirical analysis is conducted taking Beijing as an example. Compared to historical data in Beijing, results show that snowstorms in 2010 occurred more frequently than before when the intensity was weaker. Meanwhile, the frequency of low-temperature and frost disasters occurring did not change much, but serious low-temperature disasters happened due to heavy snowfalls. The indirect economic losses of the snow, low-temperature and frost disasters occurring in Beijing in 2010 are 2.23 times the direct economic losses. And some sectors are very sensitive to the disasters. Furthermore, it is found that the total economic losses obtained by a dynamic input-output model are less than that evaluated by a static model. [ABSTRACT FROM AUTHOR]

A unique dimension of employee efficiency is reported in this paper. Sexuality in the work place, as well as rendezvous and romance in this context, have already received some attention from researchers but the topic is underexplored, most probably due to differences between Western culture and Asian culture. The primary focus of the study was to investigate the role played by religion in determining an employee's sexual behavior, and in asking them to act within certain limits? The socio-economic circumstances, norms, customs, and obsession with social status that lead to delayed marriages push people into a corner, leaving them with the possible option of religiously forbidden actions where sexual desire is concerned. After indulging in such actions, regret and remorse create sexual frustration in employees, and this sexual frustration compromises their efficiency in the workplace. Ordinal regression and manipulations in syntax of ordinal regression were used to measure this concept. The study revealed that religiously forbidden actions and sexual frustration significantly diminishes the efficiency of some employees, whereas married and female employees are less sexually frustrated and thus are more efficient in the workplace. [ABSTRACT FROM AUTHOR]

We extend LS-SVM to ordinal regression, which has wide applications in many domains such as social science and information retrieval where human-generated data play an important role. Most current methods based on SVM for ordinal regression suffer from the problem of ignoring the distribution information reflected by the samples clustered around the centers of each class. This problem would degrade the performance of SVM-based methods since the classifiers only depend on the scattered samples on the border which induce large margin. Our method takes the samples clustered around class centers into account and has a competitive computational complexity. Moreover, our method would easily produce the optimal cut-points according to the prior class probabilities and hence may obtain more reasonable results when the prior class probabilities are not the same. Experiments on simulated datasets and benchmark datasets, especially on the real ordinal datasets, demonstrate the effectiveness of our method. [ABSTRACT FROM AUTHOR]

To alleviate the traffic congestion on urban expressway, Variable Message Signs (VMS) have been widely employed for diverting some traffic to arterial streets in large cities. The diversion rate is a key parameter to design, operation and evaluation of the VMS. Many works have been done to study the diversion rate by stated preference survey. However, no consistent conclusion has been obtained on the key factors influencing diversion behavior. Therefore, this paper pays attention to analyzing the drivers' diversion from expressway behavior under different network and traffic conditions, and finding the key factors influencing diversion behavior by using stated preference survey. In the survey, graphical traffic information is provided as scenario simulation. Ordinal regression method is adopted to explore the key factors influencing diversion willingness. Binary logit model is used to explore the divert decision with given traffic information. It is shown that driving years, income, driving style, average trip length, and degree of trust in VMS are the significant factors to diversion willingness. Furthermore, the same VMS displaying state results in different diversion flow at different sites. Those findings are the foundation to optimize traffic information releasing strategy and improve the road network efficiency. [ABSTRACT FROM AUTHOR]

One of the main objectives of the common agricultural policy (CAP) is the increase of productivity of agricultural units. CAP is set and reformed in the frame of providing financial support to producers in primary sector via a single payment schema as a direct funding based on the land area, where farmers obtain 'entitlements'. In Greece the total area is taken into account to the financial aid as form of direct payment. Therefore, the production is significantly decreased. In this paper the evaluation of agricultural units is proposed using the robust ordinal regression approach, and a case study dealing with the evaluation of Greek farmers in the industry of the juicing citrus is conducted. The method is used as an evaluation tool for financial aid to the farmers, towards the new policy, thus making the production-based approach more effective and objective. An additive evaluation model is proposed based on a consistent family of criteria composed by input and output criteria. The phenomenon of 'Sofa Farmers' could be eliminated, since farmers would be financially aided after been evaluated. Finally, in order to obtain robust conclusions, post-optimality analyses are applied by computing complementary robustness measures. [ABSTRACT FROM AUTHOR]

Predatory species' usage of different prey types is affected by both prey availability and selectivity. The diet during the breeding season may affect the reproductive success of individual pairs. We studied the prey use of a small reversed size-dimorphic raptor, the Eurasian sparrowhawk Accipiter nisus, with respect to prey weight on two organizational levels. Using 13 years of data from southern Norway, we related reproductive output of individual breeding events to prey size taken. Further, we assessed the regional variation in prey usage between five Fennoscandian populations. This was done by fitting optimum-type functions to the prey species' numbers or relative predation risks. Pairs that successfully completed the season with more fledglings displayed less variation in prey size, suggesting a possible adaptive benefit of diet specialism, or possibly a correlative effect due to higher prey availability or lower female hunting effort. This finding contrasts with earlier raptor studies, which have suggested benefits of dietary (and hence nutritional) diversity. Indeed, our results might be limited to nutritionally substitutable prey items. We also found a tendency suggesting that older females raised more fledglings than 1-year-old females. In the population-level analysis, we found that optimum-type functions with constant width and spatially variable average best described the relationship between relative predation risk and log weight. This can reflect local conditions, such as prey availability. Our findings and new methodological tools could apply to a broader spectrum of predators. They also highlight the role of viewing usage or choice of prey at several spatial scales. [ABSTRACT FROM AUTHOR]

This paper presents a detailed study of tsunami hazard in ports and its correlation with the damage suffered by marine vessels. The study aims to develop a new loss function to estimate the potential damage of marine vessels due to tsunami attack based on a novel multivariate statistical modeling method, which used several explanatory variables simultaneously to estimate an outcome or the probability of such outcome. In the first part of the paper, tsunami heights and velocities are numerically modeled by using high-resolution bathymetry and topography data for the southern part of Honshu Island. We apply statistical methods to the complete sequence of spatially distributed time series of tsunami parameters in order to obtain the best fit with the observed damage data. In the second part, we develop loss functions for marine vessels by using ordinal regression, which uses simultaneously several explanatory variables. We perform several statistical tests to determine the appropriate model variables to be used in developing three-dimensional loss estimation surfaces, which provide probability of loss for each combination of measured or simulated values of tsunami parameters. The main feature of the developed loss functions presented in this study is their capability to integrate the key factors influencing the damage probability, such as tsunami parameters, characteristics of marine vessels and the impact of collision experienced by the vessels during the tsunami. Such a robust method, therefore, is crucially important to understand the tsunami impact on ports and, particularly on marine vessels. [ABSTRACT FROM AUTHOR]

Ordinal regression is to predict categories of ordinal scale and it has wide applications in many domains where the human evaluation plays a major role. So far several algorithms have been proposed to tackle ordinal regression problems from a machine learning perspective. However, most of these algorithms only seek one direction where the projected samples are well ranked. So a common shortcoming of these algorithms is that only one dimension in the sample space is used, which would definitely lose some useful information in its orthogonal subspaces. In this paper, we propose a novel ordinal regression strategy which consists of two stages: firstly orthogonal feature vectors are extracted and then these projector vectors are combined to learn an ordinal regression rule. Compared with previous ordinal regression methods, the proposed strategy can extract multiple features from the original data space. So the performance of ordinal regression could be improved because more information of the data is used. The experimental results on both benchmark and real datasets proves the performance of the proposed method. [ABSTRACT FROM AUTHOR]

Tsunamis are disastrous events typically causing loss of life, and extreme damage to the built environment, as shown by the recent disaster that struck the East coast of Japan in 2011. In order to quantitatively estimate damage in tsunami prone areas, some studies used a probabilistic approach and derived fragility functions. However, the models chosen do not provide a statistically sound representation of the data. This study applies advanced statistical methods in order to address these limitations. The area of study is the city of Ishinomaki in Japan, the worst affected area during the 2011 event and for which an extensive amount of detailed building damage data has been collected. Ishinomaki city displays a variety of geographical environments that would have significantly affected tsunami flow characteristics, namely a plain, a narrow coast backed up by high topography (terrain), and a river. The fragility analysis assesses the relative structural vulnerability between these areas, and reveals that the buildings surrounding the river were less likely to be damaged. The damage probabilities for the terrain area (with relatively higher flow depths and velocities) were lower or similar to the plain, which confirms the beneficial role of coastal protection. The model diagnostics show tsunami flow depth alone is a poor predictor of tsunami damage for reinforced concrete and steel structures, and for all structures other variables are influential and need to be taken into account in order to improve fragility estimations. In particular, evidence shows debris impact contributed to at least a significant amount of non-structural damage. [ABSTRACT FROM AUTHOR]

Based on the classification provided by the Ministry of Land, Infrastructure, Transport and Tourism (MLIT), the damage level of buildings impacted by the 2011 Great East Japan tsunami can be separated into six levels (from minor damage to washed away). The objective of this paper is to identify the significant predictor variables and the direction of their potential relationship to the damage level in order to create a predicting formula for damage level. This study used the detailed data of damaged buildings in Ishinomaki city, Miyagi prefecture, Japan, collected by MLIT. The explanatory variables tested included the inundation depth, number of floors, structural material, and function of the building. Ordinal regression was applied to model the relationship between the ordinal outcome variable (damage level) and the predictors. The findings indicated that inundation depth, structural material, and function of building were significantly associated with the damage level. In addition to this new type of model, this research provides a valuable insight into the relative influence of different factors on building damage and suggestions that may help to revise the classification of current standards. This study can contribute to academic tsunami research by assessing the contribution of different variables to the observed damage using new approaches based on statistical analysis and regression. Moreover, practical applications of these results include understanding of the predominant factors driving tsunami damage to structures, implementation of the relevant variables into the proposed, or alternative model in order to improve current damage predictions by taking into account not only inundation depth, but also variables such as structural material and function of building. [ABSTRACT FROM AUTHOR]

Tsunamis are destructive natural phenomena which cause extensive damage to the built environment, affecting the livelihoods and economy of the impacted nations. This has been demonstrated by the tragic events of the Indian Ocean tsunami in 2004, or the Great East Japan tsunami in 2011. Following such events, a few studies have attempted to assess the fragility of the existing building inventory by constructing empirical stochastic functions, which relate the damage to a measure of tsunami intensity. However, these studies typically fit a linear statistical model to the available damage data, which are aggregated in bins of similar levels of tsunami intensity. This procedure, however, cannot deal well with aggregated data, low and high damage probabilities, nor does it result in the most realistic representation of the tsunami-induced damage. Deviating from this trend, the present study adopts the more realistic generalised linear models which address the aforementioned disadvantages. The proposed models are fitted to the damage database, containing 178,448 buildings surveyed in the aftermath of the 2011 Japanese tsunami, provided by the Ministry of Land, Infrastructure Transport and Tourism in Japan. In line with the results obtained in previous studies, the fragility curves show that wooden buildings (the dominant construction type in Japan) are the least resistant against tsunami loading. The diagnostics show that taking into account both the building's construction type and the tsunami flow depth is crucial to the quality of the damage estimation and that these two variables do not act independently. In addition, the diagnostics reveal that tsunami flow depth estimates low levels of damage reasonably well; however, it is not the most representative measure of intensity of the tsunami for high damage states (especially structural damage). Further research using disaggregated damage data and additional explanatory variables is required in order to obtain reliable model estimations of building damage probability. [ABSTRACT FROM AUTHOR]

Predicting labels of structured data such as sequences or images is a very important problem in statistical machine learning and data mining. The conditional random field (CRF) is perhaps one of the most successful approaches for structured label prediction via conditional probabilistic modeling. In such models, it is traditionally assumed that each label is a random variable from a nominal category set (e.g., class categories) where all categories are symmetric and unrelated from one another. In this paper we consider a different situation of ordinal-valued labels where each label category bears a particular meaning of preference or order. This setup fits many interesting problems/datasets for which one is interested in predicting labels that represent certain degrees of intensity or relevance. We propose a fairly intuitive and principled CRF-like model that can effectively deal with the ordinal-scale labels within an underlying correlation structure. Unlike standard log-linear CRFs, learning the proposed model incurs non-convex optimization. However, the new model can be learned accurately using efficient gradient search. We demonstrate the improved prediction performance achieved by the proposed model on several intriguing sequence/image label prediction tasks. [ABSTRACT FROM AUTHOR]

The purpose of this study was to develop an in-depth understanding of the barriers and enablers of effective dual care (care obtained from the Veterans Health Administration [VHA] and the private health system) for rural veterans. Telephone interviews of a random sample of 1,006 veterans residing in rural Nebraska were completed in 2010. A high proportion of the rural veterans interviewed reported receiving dual care. The common reasons cited for seeking care outside the VHA (or VA [Veterans Administration]) included having an established relationship with a non-VA provider and distance to the nearest VA medical center. Almost half of the veterans who reported having a personal doctor or nurse reported that this was a non-VA provider. Veterans reported high levels of satisfaction with the quality of care they receive. Ordinal logistic regression models found that veterans who were Medicare beneficiaries, and who rated their health status higher had higher satisfaction with dual care. The reasons cited by the veterans for seeking care at the VHA (quality of VHA care, lower costs of VHA care, entitlement) and veterans perceptions about dual care (confused about where to seek care for different ailments, perceived lack of coordination between VA and non VA providers) were significant predictors of veterans' satisfaction with dual care. This study will guide policymakers in the VA to design a shared care system that can provide seamless, timely, high quality and veteran centered care. [ABSTRACT FROM AUTHOR]

Online product reviews are increasingly being recognized as a gold mine of information for determining product and brand positioning, and more and more companies look for ways of digging this gold mine for nuggets of knowledge that they can then bring to bear in decision making. We present a software system, called StarTrack, that automatically rates a product review according to a number of 'stars,' i.e., according to how positive it is. In other words, given a text-only review (i.e., one with no explicit star-rating attached), StarTrack attempts to guess the star-rating that the reviewer would have attached to the review. StarTrack is thus useful for analysing unstructured word-of-mouth on products, such as the comments and reviews about products that are to be found in spontaneous discussion forums, such as newsgroups, blogs, and the like. StarTrack is based on machine learning technology, and as such does not require any re-programming for porting it from one product domain to another. Based on the star-ratings it attributes to reviews, StarTrack can subsequently rank the products in a given set according to how favourably they have been reviewed by consumers. We present controlled experiments in which we evaluate, on two large sets of product reviews crawled from the Web, the accuracy of StarTrack at (i) star-rating reviews, and (ii) ranking the reviewed products based on the automatically attributed star-ratings. [ABSTRACT FROM AUTHOR]

This paper proposes a hierarchical probabilistic model for ordinal matrix factorization. Unlike previous approaches, we model the ordinal nature of the data and take a principled approach to incorporating priors for the hidden variables. Two algorithms are presented for inference, one based on Gibbs sampling and one based on variational Bayes. Importantly, these algorithms may be implemented in the factorization of very large matrices with missing entries. The model is evaluated on a collaborative filtering task, where users have rated a collection of movies and the system is asked to predict their ratings for other movies. The Netflix data set is used for evaluation, which consists of around 100 million ratings. Using root mean-squared error (RMSE) as an evaluation metric, results show that the suggested model outperforms alternative factorization techniques. Results also show how Gibbs sampling outperforms variational Bayes on this task, despite the large number of ratings and model parameters. Matlab implementations of the proposed algorithms are available from . [ABSTRACT FROM AUTHOR]