Search

Differences in data size per class, also known as imbalanced data distribution, have become a common problem affecting data quality. Big Data scenarios pose a new challenge to traditional imbalanced classification algorithms, since they are not prepared to work with such amount of data. Split data strategies and lack of data in the minority class due to the use of MapReduce paradigm have posed new challenges for tackling the imbalance between classes in Big Data scenarios. Ensembles have been shown to be able to successfully address imbalanced data problems. Smart Data refers to data of enough quality to achieve high-performance models. The combination of ensembles and Smart Data, achieved through Big Data preprocessing, should be a great synergy. In this paper, we propose a novel Smart Data driven Decision Trees Ensemble methodology for addressing the imbalanced classification problem in Big Data domains, namely SD_DeTE methodology. This methodology is based on the learning of different decision trees using distributed quality data for the ensemble process. This quality data is achieved by fusing random discretization, principal components analysis, and clustering-based random oversampling for obtaining different Smart Data versions of the original data. Experiments carried out in 21 binary adapted datasets have shown that our methodology outperforms random forest. [ABSTRACT FROM AUTHOR]

This paper presents, for the first time, a distance metric learning algorithm for monotonic classification. Monotonic datasets arise in many real-world applications, where there exist order relations in the input and output variables, and the outputs corresponding to ordered pairs of inputs are also expected to be ordered. Monotonic classification can be addressed through several distance-based classifiers that are able to respect the monotonicity constraints of the data. The performance of distance-based classifiers can be improved with the use of distance metric learning algorithms, which are able to find the distances that best represent the similarities among each pair of data samples. However, learning a distance for monotonic data has an additional drawback: the learned distance may negatively impact the monotonic constraints of the data. In our work, we propose a new model for learning distances that does not corrupt these constraints. This methodology will also be useful in identifying and discarding non-monotonic pairs of samples that may be present in the data due to noise. The experimental analysis conducted, supported by a Bayesian statistical testing, demonstrates that the distances obtained by the proposed method can enhance the performance of several distance-based classifiers in monotonic problems. [ABSTRACT FROM AUTHOR]

In sentiment analysis of microblogs that involve social relationships, a common approach is to expand the features of target microblogs using microblog relationship networks. However, the current research methodology only relies on individual interaction behaviors on social platforms to construct these networks, disregarding the guiding influence of microblog relationship networks on microblogs. Consequently, this leads to the feature expansion of microblogs while introducing interference among them. To address this problem, this study aims to construct a more precise microblog relationship network by incorporating multiple interactive behaviors from social platforms. This network will serve as a guide for sentiment interactions among microblog texts, thereby mitigating feature interference. Firstly, we utilize various interaction behaviors on social platforms to build a microblog relationship network. We employ a LINE network embedding to represent the microblog relationship network as microblog relationship features. Secondly, we extract word-level and sentence-level features from the microblog text using a BERT pre-training model. The word-level features are combined using convolutional neural networks. Subsequently, the word-level and sentence-level features of microblogs are separately guided for interaction fusion through relational features. An attention network is then constructed to fuse the post-interaction features in a single step. Prior to secondary fusion, the primary fusion features, post-interaction word-level features, and post-interaction sentence-level features are weighted, and the sentiment categories of microblogs are outputted. Finally, we compare the proposed method with the text-only microblog sentiment analysis approach and the sentiment analysis method that incorporates social relationships on two real datasets. The comparison results demonstrate the superiority of our proposed method. [ABSTRACT FROM AUTHOR]

Many real-world datasets exhibit imbalanced distributions, in which the majority classes have sufficient samples, whereas the minority classes often have a very small number of samples. Data resampling has proven to be effective in alleviating such imbalanced settings, while feature selection is a commonly used technique for improving classification performance. However, the joint impact of feature selection and data resampling on two-class imbalance classification has rarely been addressed before. This work investigates the performance of two opposite imbalanced classification frameworks in which feature selection is applied before or after data resampling. We conduct a large-scale empirical study with a total of 9225 experiments on 52 publicly available datasets. The results show that both frameworks should be considered for finding the best performing imbalanced classification model. We also study the impact of classifiers, the ratio between the number of majority and minority samples (IR), and the ratio between the number of samples and features (SFR) on the performance of imbalance classification. Overall, this work provides a new reference value for researchers and practitioners in imbalance learning. [ABSTRACT FROM AUTHOR]

Background : Burns are one of the most common causes of mortality and morbidity among children. This study aims to assess the epidemiology of pediatric major burns in a third level hospital in Spain to evaluate demographics, etiology, and outcomes. Methods: A retrospective study was held by the Plastic, Reconstructive and Burn Surgery department of the hospital. We included 147 patients under 18 admitted to hospital between January 2008 and December 2020 who meet the inclusion criteria: partial thickness burns > 10% total body surface area (TBSA) in patients < 18 years old. Clinical data extracted included age, gender, date of admission, %TBSA, burn types, severity and sites of burn, length of stay, length of ventilator support, intensive care admission, blood transfusion, surgical interventions, and complications. Results: Three groups of age were analyzed. The average %TBSA was 18.7 (SE 0.9). Scalds were the main mechanism of injury (70.1%) and upper extremity was the most frequent location affected (68%). The 28.6% of patients suffered some complication, but the mortality rate was low (0.7%). In our series, the group aged 13–18 showed significantly higher %TBSA, more number of surgeries and blood transfusions. Conclusions: Scald burns are the most frequent mechanism of injury in pediatric burns. However, teenagers suffer more severe burns and complications, usually caused by flame. Despite the low mortality rates, more measures of prevention should be taken to increase children security. Level of evidence: Level IV, Risk/Prognostic. [ABSTRACT FROM AUTHOR]

Background: The Burn Specific Health Scale-Brief (BSHS-B) is considered a validated questionnaire to evaluate quality of life after burn. The purpose of this study was to translate the BSHS-B into Spanish and perform its cross-cultural adaptation. Methods: First, BSHS-B was translated from English into Spanish. Subsequently, 84 patients answered the Spanish version of BSHS-B and SF-36 scales to assess the reliability (Cronbach's alpha), construct validity (Spearman rank test), and stability in time (intra-class correlation coefficient). Results: The BSHS-B-Sp showed a good internal consistency with a global Cronbach's alpha of 0.96. Correlations between the major domains of BSHS-B and SF-36 are reported with Spearman's coefficient ranging from 0.29 to 0.87 in all sub-domains (p < 0.01). Stability in time was confirmed by intra-class correlation coefficient (0.91–0.99 for every sub-domain and 0.98 for global score, p < 0.001). Conclusions: The Spanish version of the BSHS-B demonstrated an appropriated internal consistency, construct validity, and stability in time. This fact supports its applicability to evaluate the quality of life of burn patients. Level of evidence: Not gradable [ABSTRACT FROM AUTHOR]

Introduction: The optimal anti-thrombotic therapy to prevent recurrent ischemic events in patients with acute coronary syndrome and coronary artery ectasia (CAE) remains unclear. Aim: To assess the efficacy and safety of antiplatelet plus anticoagulant therapy versus dual antiplatelet therapy in patients with acute coronary syndromes and coronary artery ectasia. Methods: OVER-TIME is an investigator initiated, exploratory, open label, single center, randomized clinical trial comparing dual antiplatelet therapy (acetyl-salicylic acid plus a P2Y12 inhibitor) with the combination of an antiplatelet monotherapy (a P2Y12 inhibitor) plus a low dose anticoagulant (rivaroxaban, 15mg oral dose) for the prevention of recurrent ischemic events among patients with CAE. We aim to enroll approximately 60 patients with CAE and acute coronary syndromes. After recruitment, patients are randomized to (a) standard of care (dual antiplatelet regimen) or (b) the combination of antiplatelet monotherapy and low dose anticoagulant. Patients will be followed for at least 12 months. The OVER-TIME study aims to assess the efficacy of the regimen in prevention of major cardiovascular events and its security in bleeding events in acute coronary syndromes among patients with CAE. Expected results and conclusions: OVER-TIME is the first randomized controlled trial to assess different antithrombotic strategies in patients with CAE and acute coronary syndrome, and its results will offer preliminary data for the prevention of major cardiovascular events and bleeding events in this group of patients. Trial Registration Number: NCT05233124 (ClinicalTrials.gov), date of registration: February 10, 2022. [ABSTRACT FROM AUTHOR]

Objectives: The purpose of this exploratory non-randomized controlled study was to determine the acceptance and effectiveness of an 8-week mindfulness-based intervention (MBI) co-designed by a police officer. Methods: A pretest-posttest control group design was followed. Participants (MBI group = 20; control group = 18) answered baseline and post-training self-reported measures. In addition, the weekly emotional state of the MBI group was collected. Paired-samples t-test and analysis of covariance were performed for pre-post within-group and between-group differences, respectively, as well as linear mixed effects analysis of repeated measures for week-by-week data. Results: High acceptance and attendance rates, as well as significant pre-post within-group differences in the MBI group in mindfulness (η2 = 0.43), self-compassion (η2 = 0.43), depression (η2 = 0.54), anxiety (η2 = 0.46), stress (η2 = 0.51), difficulties in emotion regulation, sleep quality (η2 = 0.57), and burnout (η2 = 0.31–0.47), were identified. Moreover, police officers who underwent the MBI experienced a week by week decrease of anger, disgust, anxiety, sadness, and desire. Finally, after adjusting for pre-test scores, significant between-group differences were found in the way of attending to internal and external experiences (observing mindfulness facet; ηp2 = 0.21), depression symptoms (ηp2 = 0.23), general distress (ηp2 = 0.24), and the degree of physical and psychological exhaustion (personal burnout; ηp2 = 0.20). Conclusions: The preliminary effectiveness of this MBI on psychopathology and quality of life outcomes in Spanish police officers was discussed. Previous evidence regarding the promising use of MBIs in this population was supported. [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]

In recent algorithmic family simulates different biological processes observed in Nature in order to efficiently address complex optimization problems. In the last years the number of bio-inspired optimization approaches in literature has grown considerably, reaching unprecedented levels that dark the future prospects of this field of research. This paper addresses this problem by proposing two comprehensive, principle-based taxonomies that allow researchers to organize existing and future algorithmic developments into well-defined categories, considering two different criteria: the source of inspiration and the behavior of each algorithm. Using these taxonomies we review more than three hundred publications dealing with nature-inspired and bio-inspired algorithms, and proposals falling within each of these categories are examined, leading to a critical summary of design trends and similarities between them, and the identification of the most similar classical algorithm for each reviewed paper. From our analysis we conclude that a poor relationship is often found between the natural inspiration of an algorithm and its behavior. Furthermore, similarities in terms of behavior between different algorithms are greater than what is claimed in their public disclosure: specifically, we show that more than one-third of the reviewed bio-inspired solvers are versions of classical algorithms. Grounded on the conclusions of our critical analysis, we give several recommendations and points of improvement for better methodological practices in this active and growing research field. [ABSTRACT FROM AUTHOR]

Previous studies show that traditional teaching methods such as oral explanations and PowerPoint presentations can be complemented with the use of computer simulations in problem-solving sessions to teach robotics. Nevertheless, these previous works rest upon the assumption that the knowledge that is learned in virtual laboratories transfers to its equivalent real task. The main contribution of this work is to validate that the knowledge obtained by students in an undergraduate robotics course using computer simulations of 3D model robots can be applied to its real-world context. The experimental platform used to run the computer simulations was based on the Simscape Multibody library of the MATLAB™ software. Results show a satisfactory knowledge transfer because more than 75% of the students finished a forward kinematics problem using a physical manipulator built with LEGO Mindstorms™. In addition, the observation analysis of the professor reveals that students reinforced their knowledge previously learned, and their problem-solving and critical thinking skills also improved. In this way, this study demonstrates that the proposed methodology not only helps the transfer of knowledge to the real-world context but also helps to develop student competencies. Future work is needed to replicate the findings in other robotics topics, such as the generation of trajectories, the analysis of the dynamics of manipulators, and the design of various controllers for the actuators of the robot. [ABSTRACT FROM AUTHOR]

In the educational area of engineering, there has been research using simulations to illustrate the concepts and exercises seen in class, as in the case of robotics course classes in which simulations have been used to demonstrate theoretical concepts. Results have been inconclusive whether or not the use of state-of-the-art simulations with traditional teaching methods improve the academic performance of students. From the point of view of the authors, some of these negative results could be because teachers do not receive any guidance on how to incorporate the use of simulations in their teaching strategy. This paper reports the evaluation of a platform to simulate industrial robots based on the MATLAB™ and Simulink™ environment. The platform was used by 3 teachers of undergraduate robotics courses. Each teacher was trained how to use the platform. For teachers of groups A and B, in addition to this training, they also received guidance on how to use the platform in their classes, while the teacher of group C freely decided on his own how to use the platform in his classes. In each group, the teacher covered the same topics about forward kinematics, and the same exercises were given to the students to solve during the practical sessions. As a result, students in groups A and B attained better academic performance compared to students of group C. This fact shows that it is necessary to train the professors not only in the use of the platform but also in the incorporation of this technology into their courses. [ABSTRACT FROM AUTHOR]

Instance reduction techniques are data preprocessing methods originally developed to enhance the nearest neighbor rule for standard classification. They reduce the training data by selecting or generating representative examples of a given problem. These algorithms have been designed and widely analyzed in multi-class problems providing very competitive results. However, this issue was rarely addressed in the context of one-class classification. In this specific domain a reduction of the training set may not only decrease the classification time and classifier's complexity, but also allows us to handle internal noisy data and simplify the data description boundary. We propose two methods for achieving this goal. The first one is a flexible framework that adjusts any instance reduction method to one-class scenario by introduction of meaningful artificial outliers. The second one is a novel modification of evolutionary instance reduction technique that is based on differential evolution and uses consistency measure for model evaluation in filter or wrapper modes. It is a powerful native one-class solution that does not require an access to counterexamples. Both of the proposed algorithms can be applied to any type of one-class classifier. On the basis of extensive computational experiments, we show that the proposed methods are highly efficient techniques to reduce the complexity and improve the classification performance in one-class scenarios. [ABSTRACT FROM AUTHOR]

The aim of the present work is to validate a model of antecedents and consequences of university perceived value among graduates, and to analyse the moderating role of their level of involvement with Higher Education. Perceived value is proposed as the principal antecedent of perceived value, comprising four dimensions: teaching staff, infrastructures, administration staff, and support services. Overall satisfaction with the university and the overall image of the institution are taken as consequences, for the purpose of this study. The final sample - comprising 352 university graduates from all areas of knowledge - was obtained via computer-assisted telephone interviewing (CATI). The theoretical model was estimated using Consistent Partial Least Squares (PLSc) and Multi-group Analysis (PLS-MGA). The results of the analysis confirm that perceived quality is a multidimensional construct formed by four distinct dimensions. It is found to determine perceived value, which, in turn, has a major influence on overall satisfaction and overall image of the institution, although the latter depends directly on satisfaction but not on perceived value in the case of graduates with a high level of involvement. These findings contribute to a better understanding of how university management can improve graduate satisfaction, via perceived quality and perceived value, and how graduate satisfaction is a major antecedent of overall image-formation. Further, the present results confirm that the extent of graduates' involvement in Higher Education appears to moderate the effect of value on image, this moderation being greater among those graduates presenting a lower level of involvement. The conclusions from the research provide valuable information that can help guide university management in decision-making and strategy development. [ABSTRACT FROM AUTHOR]

Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. The online version of the original article can be found at https://doi.org/10.1007/s10489-022-03772-1 B Correction to: Applied Intelligence b https://doi.org/10.1007/s10489-022-03772-1 The original online version of this article was revised due to incorrect captions and images of Figures 10, 11, 12 and 13. [Extracted from the article]

In supervised prediction problems, the response attribute depends on certain explanatory attributes. Some real problems require the response attribute to represent ordinal values that should increase with some of the explaining attributes. They are called classification problems with monotonicity constraints. In this paper, we aim at formalizing the approach to nested generalized exemplar learning with monotonicity constraints, proposing the monotonic nested generalized exemplar learning (MoNGEL) method. It accomplishes learning by storing objects in $${\mathbb {R}}^n$$ , hybridizing instance-based learning and rule learning into a combined model. An experimental analysis is carried out over a wide range of monotonic data sets. The results obtained have been verified by non-parametric statistical tests and show that MoNGEL outperforms well-known techniques for monotonic classification, such as ordinal learning model, ordinal stochastic dominance learner and k-nearest neighbor, considering accuracy, mean absolute error and simplicity of constructed models. [ABSTRACT FROM AUTHOR]

Audio fingerprinting allows us to label an unidentified music fragment within a previously generated database. The use of spectral landmarks aims to obtain a robustness that lets a certain level of noise be present in the audio query. This group of audio identification algorithms holds several configuration parameters whose values are usually chosen based upon the researcher's knowledge, previous published experimentation or just trial and error methods. In this paper we describe the whole optimisation process of a Landmark-based Music Recognition System using genetic algorithms. We define the actual structure of the algorithm as a chromosome by transforming its high relevant parameters into various genes and building up an appropriate fitness evaluation method. The optimised output parameters are used to set up a complete system that is compared with a non-optimised one by designing an unbiased evaluation model. [ABSTRACT FROM AUTHOR]

There are increasing demands on universities to operate transparently with regard how resources are being used and targets met, putting them under growing pressure to clarify their position relative to other universities at national and international level. But there are several challenges associated with establishing their relative positioning. The first issue is concerned with measurement or how to capture data that are relevant, pertinent and fit-for-purpose. Second, universities must obtain an overall indicator or a means of ordering that helps synthesize the different indicators; and, third, they must decide how to weight them. Those university rankings that attempt to address these questions are met with a degree of criticism for being subjective or inconsistent in their quantification of the indicators. The aim of the present work is to develop a procedure for synthesizing all of the indicators relating to the objective measurement of university R&D and innovation into a single or summary concept. In other words, to establish a procedure that does not require subjective criteria and that can be applied for both absolute and relativized indicators. This approach makes a dual contribution. First, a specific application, in this particular case for the case of the Spanish university system, will be created, to obtain a synthetic indicator for research activity. This will enable us to achieve an R&D and innovation ranking for Spanish universities. Second, the work makes a methodological contribution by using a new technique for synthesizing this type of indicator, namely Partial Least Squares (PLS). [ABSTRACT FROM AUTHOR]