Search

Logistic regression is a standard procedure for real-world classification problems. The challenge of class imbalance arises in two-class classification problems when the minority class is observed much less than the majority class. This characteristic is endemic in many domains. Work by Owen [2007] has shown that cluster structure among the minority class may be a specific problem in highly imbalanced logistic regression. In this paper, we propose a novel relabeling approach to handle the class imbalance problem when using logistic regression, which essentially assigns new labels to the minority class observations. An Expectation-Maximization algorithm is formalized to serve as a tool for efficiently computing this relabeling. Modeling on such relabeled data can lead to improved predictive performance. We demonstrate the effectiveness of this approach with detailed experiments on real data sets.

Objective This study examined whether professional air traffic controllers (ATCos) were subject to peak-end effects in reporting their mental workload after performing an air traffic control task, and in predicting their mental workload in future scenarios. Background In affective experience studies, people’s evaluation of a period of experience is strongly influenced by the most intense (peak) point and the endpoint. However, whether the effects exist in mental workload evaluations made by professional operators is still not known. Method In Study 1, 20 ATCos performed air traffic control scenarios on high-fidelity radar simulators and reported their mental workload. We used a 2 (high peak, low peak) × 2 (high end, low end) within-subject design. In Study 2, another group of 43 ATCos completed a survey asking them to predict their mental workload given the same air traffic control scenarios. Results In Study 1, ATCos reported higher mental workload after completing the high-peak and the high-end scenarios. In contrast, in Study 2, ATCos predicted the peak workload effect but not the end workload effect when asked to predict their experience in dealing with the same scenarios. Conclusion Peak and end effects exist in subjective mental workload evaluation, but experts only had meta-cognitive awareness of the peak effect, and not the end effect. Application Researchers and practitioners that use subjective workload estimates for work design decisions need to be aware of the potential impact of peak and end task demand effects on subjective mental workload ratings provided by expert operators.

Mesh-based numerical methods such as the Finite Element Method (FEM) and Finite Difference Method (FDM) are widely used in various fields of science and engineering. However, the shortcomings of mesh-based methods are becoming increasingly obvious mainly due to the distorted or broken elements when analyzing the problems of large deformation or crack propagation. To overcome the above problems, meshfree methods such as the Meshless Local Petrov–Galerkin (MLPG) and Radial Point Interpolation Method (RPIM) were proposed, which can eliminate or partially eliminate the difficulties triggered by mesh-dependence. In geotechnics, the numerical modeling of deformations and failures of soil and rock masses is one of the most important approaches to understand and prevent geohazards, such as landslides and ground subsidence. In this paper, a local RPIM-based meshfree software package GeoMFree 3D is specifically developed for analyzing the deformations and failures of soil and rock masses. To enhance the computing capability of GeoMFree 3D , (1) effective memory storage strategies are adopted to solve the memory requirement bottleneck; and (2) parallel computing strategies are utilized and heterogeneously implemented on multi-core CPU and many-core GPU to improve the computational efficiency. To demonstrate the performance of GeoMFree 3D , three simple verification examples and one application case are presented in this paper. The verification examples indicate that the computational accuracy and efficiency of the package GeoMFree 3D is acceptable. Furthermore, the application case indicates that GeoMFree 3D is capable of analyzing the displacements of complex rock slopes. GeoMFree 3D is currently under intensive development, plans for future work on the development of GeoMFree 3D are also introduced.

The flow-difference feedback iteration method for aerostatic thrust bearings is proposed to the flow balance-based iteration, and two modification methods are further provided to improve the adaptability. The bearing capability calculated by the proposed method is validated by the experimental data. Moreover, the influence of convergence rate factors, iterative initial values, and mesh grids on the iteration ratio is investigated. Compared with the conventional iteration methods, the proposed method with appropriate convergence rate factors provides a higher convergence efficiency. In addition, good convergence behavior under different iterative initial values and the mesh grid size is shown, and the convergence rate is insensitive to the finite difference method parameters. A series of calculations are conducted to investigate the generality of the proposed methods.

The class imbalance problem arises in two-class classification problems, when the less frequent (minority) class is observed much less than the majority class. This characteristic is endemic in many problems such as modeling default or fraud detection. Recent work by Owen [ 19 ] has shown that, in a theoretical context related to infinite imbalance, logistic regression behaves in such a way that all data in the rare class can be replaced by their mean vector to achieve the same coefficient estimates. We build on Owen's results to show the phenomenon remains true for both weighted and penalized likelihood methods. Such results suggest that problems may occur if there is structure within the rare class that is not captured by the mean vector. We demonstrate this problem and suggest a relabelling solution based on clustering the minority class. In a simulation and a real mortgage dataset, we show that logistic regression is not able to provide the best out-of-sample predictive performance and that an approach that is able to model underlying structure in the minority class is often superior.

People’ retrospective evaluation of their hedonic experience is heavily influenced by the most intense (peak) and the last (end) moment. This study examines whether such peak-end effect can also be applied to the evaluation of mental workload in the context of air traffic control. Twenty professional approach controllers finished four tasks on high fidelity simulators and rated overall workload after each task by NASA-TLX scale. A 2 (high vs. low peak) by 2 (high vs. low end) within-subject design was used. The results showed that there was a significant peak effect and a marginally significant end effect. This study calls for further investigation of the measurement issue of mental workload and provides insights for workload management strategies.

In order to efficiently transmit and store speech signals, speech codecs create a minimally redundant representation of the input signal which is then decoded at the receiver with the best possible perceptual quality. In this work we demonstrate that a neural network architecture based on VQ-VAE with a WaveNet decoder can be used to perform very low bit-rate speech coding with high reconstruction quality. A prosody-transparent and speaker-independent model trained on the LibriSpeech corpus coding audio at 1.6 kbps exhibits perceptual quality which is around halfway between the MELP codec at 2.4 kbps and AMR-WB codec at 23.05 kbps. In addition, when training on high-quality recorded speech with the test speaker included in the training set, a model coding speech at 1.6 kbps produces output of similar perceptual quality to that generated by AMR-WB at 23.05 kbps., ICASSP 2019

The modelling and understanding of crack propagation for elastic–plastic materials is critical in various engineering applications, such as safety analysis of concrete structures and stability analysis of rock slopes. In this paper, the local radial basis point interpolation method (LRPIM) combined with elastic–plastic theory and fracture mechanics is employed to analyse crack propagation in elastic–plastic materials. Crack propagation in elastic–plastic materials is compared using the LRPIM and eXtended finite-element method (XFEM). The comparative investigation indicates that: (i) the LRPIM results are close to the model test results, which indicates that it is feasible for analysing the crack growth of elastic–plastic materials; (ii) compared with the LRPIM, the XFEM results are closer to the experimental results, showing that the XFEM has higher accuracy and computational efficiency; and (iii) compared with the XFEM, when the LRPIM method is used to analyse crack propagation, the propagation path is not smooth enough, which can be explained as the crack tip stress and strain not being accurate enough and still needing further improvement.

An in-flight excavation centrifuge model test was conducted to investigate the behavior of a loess slope under different excavation slope angles. The displacements of the loess slope were measured using the particle image velocimetry system. Measurement results showed that the slope can be divided into several zones according to the features of the horizontal displacement. The excavation slope angles significantly affected the deformation process and the division of the slope. The term “slip zone” that described the area where strain highly concentrated when the slip surface ran through was used to analyze the failure process. The analysis results showed that the shear failure initially occurred near the excavation zone and progressed upwards and was accompanied with the tensile failure generated in the upper part of the slope. The shear failure eventually ran through the slip zone when it reached the upper tensile failure. The results of the earth pressure measurements demonstrated that the slope under excavation had undergone a stress path similar to the reduced triaxial compression tests prior to failure. It is clear that loess slopes are more likely to failure when they are influenced simultaneously by excavation and precipitation.

Inter-sector cooperation between air traffic controllers (ATCos) can provide an effective way to manage controllers’ workload by redistributing individual task-load at a group level, yet this area has not been fully explored. Based on our previous studies which have identified the effects of certain task-level features, this present research aimed to identify the influence of individual difference variables on controllers’ backup decisions. Forty licensed controllers performed thirty-two simulated final approach scenarios in which they had to decide whether to accept a hand-over request made by a controller working in the neighboring sector. We manipulated three key task-level features across scenarios: (1) participants’ task-load, (2) the requestors’ task-load and (3) the close-landing demands of the to-be-hand-over aircraft. We also measured controllers’ work experience and job satisfaction. HLM analysis showed that: after controlling for the effects of task-level variables, job satisfaction had a unique contribution and also interacted with task-level variables in predicting backing up behaviors.

This paper describes the implementation of four-dimensional building information modelling and its associated visualisation benefits during construction of the £1 billion Tottenham Court Road station redevelopment project in central London. It explains how a construction model was developed from design models and then linked to the construction programme; and how the model was subsequently maintained during the construction stage and shared within the project team for coordination, planning, logistics, phasing and studies for specific areas.

The sliding distance is the most critical risk factors in landslide disasters, and many factors affects landslide sliding distance. In order to investigate the influence of various factors on the landslide sliding distance, take the loess landslides on south of Jing he River in Xi'an as the study area, make use of the discrete element simulation software PFC2D on the soil of the study area, establish a model take the five factors of the friction coefficient, the sphere particle size, slope, volume, height as the model variables, and analysis the influence of these five factors on the landslide sliding distance. By regression analysis, a number of numerical results can fit a sliding distance prediction equation of landslides, summarize the sliding distance landslide weights of various factors on the impact of the sliding distance. Keywords-landslide; sliding distance prediction; discrete element; linear regression analysis

A general numerical approach is developed to simulate the failure process of structure or heterogeneous material, in which different local failure models are defined for different structures and materials to describe the failure of structure joint or material element. A two-parameter Weibull distribution is employed to produce the initial heterogeneity of material, i.e., the Young modulus and the failure strain. An appropriate elastic-plastic constitutive relation with failure strain value is used to describe the material behavior and failure in elements.