Unfolding Model Research Articles

MRF based image stitching of electroplated diamond wire saw: For improving visual inspection accuracy of manufacturing quality

Electroplated diamond wire saw has become the most important tool for wafer slicing in the photovoltaic field. The distribution state of abrasives on the surface of the wire saw is an important index to evaluate its manufacturing quality. However, it is difficult to inspect the abrasive distribution state along the circumferential direction on the surface of a wire saw, which is mainly limited by the acquisition of the full-cylindrical surface image of the wire saw. This work proposed a stitching method of the wire saw semi-cylindrical surface image. The unfolding model of the wire saw semi-cylindrical surface image is first established. Taking the center of gravity of the abrasives as the feature point, the matching criterion is established according to the spatial position relationship between the cameras. Based on markov random filed theory, the effective screening of feature points is realized by introducing the context information between feature points. The wire saw image is registered by the affine transformation model, and the fusion of stitching seams is realized by Laplacian image fusion algorithm. The full cylindrical surface image of an electroplated diamond wire saw can be obtained by the proposed method. Experimental results show that the time consumption of electroplated diamond wire saw image stitching is about 0.5 s, which can meet the real-time requirements of wire saw visual inspection. This work could be applied to the online inspection of the abrasive distribution state and the real-time adjustment of process parameters, which will help to improve the manufacturing quality of the electroplated diamond wire saw.

MD3: Model-Driven Deep Remotely Sensed Image Denoising

Remotely sensed images degraded by additive white Gaussian noise (AWGN) have low-level vision, resulting in a poor analysis of their contents. To reduce AWGN, two types of denoising strategies, sparse-coding-model-based and deep-neural-network-based (DNN), are commonly utilized, which have their respective merits and drawbacks. For example, the former pursue enjoyable performance with a high computational burden, while the latter have powerful capacity in completing a specified task efficiently, but this limits their application range. To combine their merits for improving performance efficiently, this paper proposes a model-driven deep denoising (MD3) scheme. To solve the MD3 model, we first decomposed it into several subproblems by the alternating direction method of multipliers (ADMM). Then, the denoising subproblems are replaced by different learnable denoisers, which are plugged into the unfolded MD3 model to efficiently produce a stable solution. Both quantitative and qualitative results validate that the proposed MD3 approach is effective and efficient, while it has a more powerful ability in generating enjoyable denoising performance and preserving rich textures than other advanced methods.

Who Considers Leaving a Job in a Pandemic?: Determinants of Online Job Search Behavior in COVID-19

Drawing from the unfolding model of voluntary turnover and behavioral immune system theory, we proposed and tested a theoretical model of how the perceived severity of a pandemic affects employees’ turnover intention as well as their virtual job-search behaviors. In an analysis of matched survey and mobile application usage data, we found that the perceived severity of COVID-19 is positively related to employees’ turnover intention, which is, in turn, positively related to the usage of mobile job-search apps. Our findings further disclosed that these relationships were weakened when employees perceived their organizations as more adaptive to COVID-19. When employees perceived their coworkers as more supportive in coping with COVID-19, however, the relationships were strengthened. Theoretical and practical implications of these findings are discussed.

Deep Unfolding Network for PAPR Reduction in Multicarrier OFDM Systems

In this letter, we propose a Deep Unfolding Network called PR-DUN to reduce the peak-to-average power ratio (PAPR), which is a thorny problem in Orthogonal Frequency-Division Multiplexing (OFDM) systems. The proposed multi-layer model is constructed by unrolling an iterative algorithm resulting in layers with trainable parameters, which are optimized to minimize a loss function related to the PAPR value. The deep unfolding model uses the backpropagation algorithm to transfer gradients backwards to adjust parameters. Furthermore, the proposed scheme can accommodate any transmit power constraint, and therefore can control the power increase caused by the auxiliary signal. Simulation results show that the proposed PR-DUN model achieves a larger PAPR reduction and a smaller bit error rate while being less computationally intensive than related solutions.

Using unfolding models to identify targeted feedback strategies for student writing

Formative assessments are used to identify student strengths and weaknesses, but they frequently do not identify targeted feedback strategies for improving student achievement. Unfolding models offer an alternative measurement paradigm for connecting recommended feedback strategies with different levels of student achievement. In this study, a focus group of English teachers identified feedback strategies for the writing of middle school students. A questionnaire was constructed based on their identification of 10 feedback strategies related to organization, development and language usage. Next, classroom teachers responded to the feedback questionnaire based on a set of 20 essays selected to vary in writing quality. An unfolding model was used to analyze teacher responses and to identify recommended feedback strategies for each essay. Findings suggest that recommended feedback strategies can be used to define zones of targeted feedback for identifying optimal feedback strategies that can be used for improving student writing at different levels of achievement.

Deep Model-Based Architectures for Inverse Problems Under Mismatched Priors

There is a growing interest in deep model-based architectures (DMBAs) for solving imaging inverse problems by combining physical measurement models and learned image priors specified using convolutional neural nets (CNNs). For example, well-known frameworks for systematically designing DMBAs include plug-and-play priors (PnP), deep unfolding (DU), and deep equilibrium models (DEQ). While the empirical performance and theoretical properties of DMBAs have been widely investigated, the existing work in the area has primarily focused on their performance when the desired image prior is known exactly. This work addresses the gap in the prior work by providing new theoretical and numerical insights into DMBAs under mismatched CNN priors. Mismatched priors arise naturally when there is a distribution shift between training and testing data, for example, due to test images being from a different distribution than images used for training the CNN prior. They also arise when the CNN prior used for inference is an approximation of some desired statistical estimator (MAP or MMSE). Our theoretical analysis provides explicit error bounds on the solution due to the mismatched CNN priors under a set of clearly specified assumptions. Our numerical results compare the empirical performance of DMBAs under realistic distribution shifts and approximate statistical estimators.

Abusive Supervision and Turnover Intentions: A Mediation-Moderation Perspective

This study intended to provide and test a unique model describing how abusive supervision increases workers’ turnover intentions, with a mediating role of emotional exhaustion and moderating role of self-esteem. We argued that emotional exhaustion exacerbates the association between abusive supervision and turnover intentions of the workers, while self-esteem buffers this relationship, based on the unfolding model of voluntary turnover as an overarching theory. The study design reflected that abusive supervision and turnover intentions are mediated by emotional exhaustion, while the relationship between abusive supervision and emotional exhaustion is further moderated by self-esteem. A well-structured and self-administered questionnaire was used to collect data from 290 respondents. The data were analyzed, and hypotheses were tested using PLS-SEM. The study findings confirmed that there exists a positive and significant link between abusive supervision and turnover intentions through the indirect effect of emotional exhaustion. On the other hand, the findings regarding moderating effect indicates that self-esteem has a significant impact among abusive supervision and emotional exhaustion. This research identified a feasible way for supervisors to grasp how diverse the responses of various workers may be using the unfolding model. These research findings have important academic and practical implications for government representatives, policymakers, and entrepreneurial educational institutes that can use these findings.

Comparison of parameter estimation approaches for multi-unidimensional pairwise preference tests.

Multidimensional forced-choice (MFC) testing has been proposed as a way of reducing response biases in noncognitive measurement. Although early item response theory (IRT) research focused on illustrating that person parameter estimates with normative properties could be obtained using various MFC models and formats, more recent attention has been devoted to exploring the processes involved in test construction and how that influences MFC scores. This research compared two approaches for estimating multi-unidimensional pairwise preference model (MUPP; Stark et al., 2005) parameters based on the generalized graded unfolding model (GGUM; Roberts et al., 2000). More specifically, we compared the efficacy of statement and person parameter estimation based on a "two-step" process, developed by Stark et al. (2005), with a more recently developed "direct" estimation approach (Lee et al., 2019) in a Monte Carlo study that also manipulated test length, test dimensionality, sample size, and the correlations between generating person parameters for each dimension. Results indicated that the two approaches had similar scoring accuracy, although the two-step approach had better statement parameter recovery than the direct approach. Limitations, implications for MFC test construction and scoring, and recommendations for future MFC research and practice are discussed.

Deep unfolding architecture for MRI reconstruction enhanced by adaptive noise maps

Unfolding provides a potent method to improve deep network performance in image restoration problems. Recent results in the literature have demonstrated the improvement achieved by unfolding structures when compared to the non-unfolding singular use of a given network. Lately, unfolding models have been offered as promising solutions for the Magnetic Resonance Image (MRI) reconstruction problem. In this work we propose a novel deep unfolding structure for MR image reconstruction. We introduce an adaptive noise level parameter to the unfolding structure, inspired by the conventional iterative thresholding-based reconstruction models. The noise level parameter is calculated at each iteration using the error between the network output and the initial zero filling estimate. This new parameter is given as an additional input to the network, and it acts as an evolving regularizer for the image manipulation strength of the network over the unrolling iterations. The introduction of this adaptivity over iterations in the training step also improves the deep models’ reconstructed image quality in the inference stage. Empirical results indicate that the recommended technique can converge to better reconstruction results when compared to state-of-the-art unfolding structures devoid of such an adaptive parameter. The introduction of the additional adaptive parameter results in an incremental increase in the parameter complexity, and the required reconstruction times also stand very similar. In this study, the statistical differences between developed techniques are investigated using the one-way ANOVA method. Additionally, a t-test is used to specify the major difference between the means of the two proposed structures. These results indicate that differences in the performance metrics results are statistically significant.

Trust violation at work: Lived experiences of American, Indian, and Chinese employees

Despite decades of research establishing that trust is critical to successful collaboration, the experience of trust violation is poorly understood independent of trust repair. Furthermore, despite increasing globalization, most organizational research on trust violation is heavily Westernized. This semi-structured interview study explored subjective experiences of trust violation at work across 23 individuals from the United States, India, and China to better understand similarities and differences in the unfolding reactions to trust violations across cultures. Our inductive thematic analysis identifies some trust violation triggers common to all three nationalities (i.e., psychological contract breach, professional attack, lack of work ethic) and some triggers unique to certain nationalities (i.e., lack of acknowledgment for Indian workers; excessive monitoring and injustice perceptions for Chinese workers). Regarding reactions to trust violations, American workers emphasized a central reaction of anger, Indian workers described more varied emotional and behavioral reactions possibly reflecting cultural complexity, and Chinese workers described reactions of emotion suppression and behavioral avoidance that align with theories of face. For American and Indian workers, violations damaged both interpersonal relationships and attitudes towards one’s job, whereas for Chinese workers, violations damaged only the focal interpersonal relationship. We discuss the implications of our descriptive, nationality-specific unfolding models of trust violation for advancing cross-cultural research on trust violations at work.

Reconstruction of Unfolding Sub-Events From Social Media Posts

Event detection plays a crucial role in social media analysis, which usually concludes sub-event detection and correlation. In this article, we present a method for reconstructing the unfolding sub-event relations in terms of external expert knowledge. First, a Single Pass Clustering method is utilized to summarize massive social media posts. Second, a Label Propagation Algorithm is introduced to detect the sub-event according to the expert labeling. Third, a Word Mover’s Distance method is used to measure the correlation between the relevant sub-events. Finally, the Markov Chain Monte Carlo simulation method is presented to regenerate the popularity of social media posts. The experimental results show that the popularity dynamic of the empirical social media sub-events is consistent with the data generated by the proposed method. The evaluation of the unfolding model is 50.52% ∼ 88% higher than that of the random null model in the case of “Shanghai Tesla self-ignition incident.” This work is helpful for understanding the popularity mechanism of the unfolding events for online social media.

A general recognition theory model for identifying an ideal stimulus

A probabilistic, multidimensional model is described that accounts for sensory and hedonic ratings that are collected from the same experiment. The model combines a general recognition theory model of the sensory ratings with Coombs’ unfolding model of the hedonic ratings. The model uses sensory ratings to build a probabilistic, multidimensional representation of the sensory experiences elicited by exposure to each stimulus, and it also builds a similar representation of the hypothetical ideal stimulus in this same space. It accounts for hedonic ratings by measuring differences between the presented stimulus and the imagined ideal on each rated sensory dimension. Therefore, it provides precise estimates of the sensory qualities of the ideal on all rated sensory dimensions. The model is tested successfully against data from a new experiment.

Resign or carry-on? District and principal leadership as drivers of change in teacher turnover intentions during the COVID-19 crisis: A latent growth model examination.

During the COVID‐19 pandemic, teachers in the United States, an already at‐risk occupation group, experienced new work‐related stressors, safety concerns, and work‐life challenges, magnifying on‐going retention concerns. Integrating the crisis management literature with the unfolding model of turnover, we theorize that leader actions trigger initial employee responses but also set the stage for on‐going crisis response that influence changes in teachers' turnover intentions. We apply latent growth curve modelling to test our hypotheses based on a sample of 617 K‐12 teachers using nine waves of data, including a baseline survey at the start of the 2020–2021 school year and eight follow‐up surveys (2‐week lags) through the Fall 2020 semester. In terms of overall adaptation, teachers on average, experienced an increase in work‐life balance and a decrease in turnover intentions over the course of the semester. Results also suggest that district and school leadership provide unique and complementary resources, but leader behaviours that shape initial crisis responses do not similarly affect employee responses during crisis, contrary to theory. Instead, teachers' adaptive crisis response trajectories were triggered by continued resource provision over the semester; increasing provision of valued resources (i.e., continued refinement of safety practices) and improvements in work‐life balance prevented turnover intentions from spiralling throughout the crisis. Crisis management theory and research should continue to incorporate temporal dynamics and identify factors that contribute to crisis response trajectories, using designs and analyses that allow for examination as crises unfold in real time.

Mechanism of Osmolyte Stabilization-Destabilization of Proteins: Experimental Evidence.

In this work, we investigated the influence of stabilizing (N,N,N-trimethylglycine) and destabilizing (urea) osmolytes on the hydration spheres of biomacromolecules in folded forms (trpzip-1 peptide and hen egg white lysozyme—hewl) and unfolded protein models (glycine—GLY and N-methylglycine—NMG) by means of infrared spectroscopy. GLY and NMG were clearly limited as minimal models for unfolded proteins and should be treated with caution. We isolated the spectral share of water changed simultaneously by the biomacromolecule/model molecule and the osmolyte, which allowed us to provide unambiguous experimental arguments for the mechanism of stabilization/destabilization of proteins by osmolytes. In the case of both types of osmolytes, the decisive factor determining the equilibrium folded/unfolded state of protein was the enthalpy effect exerted on the hydration spheres of proteins in both forms. In the case of stabilizing osmolytes, enthalpy was also favored by entropy, as the unfolded state of a protein was more entropically destabilized than the folded state.

A unified framework of deep unfolding for compressed color imaging

Traditional iterative-based reconstruction algorithms for compressed color imaging often suffer from long reconstruction time and low reconstruction accuracy at extreme low-rate subsampling. This paper proposes a model-driven deep learning framework for compressed color imaging. In the training step, extract the image blocks at the same position of the R, G, and B channel images as the ground truth, and then, singular value decomposition is performed on the measurement matrix to obtain the optimized measurement matrix and low-dimensional measurements; afterward, the ground-truth and optimized measurements are utilized to construct a large amount of training data pairs to train an ‘end-to-end’ deep unfolding model for compressed color imaging. In the test step, the single pretrained model is used to reconstruct high-quality images from optimized low-dimensional compressed measurements for each channel and synthesize a color image. Numerical experiments demonstrate that our proposed unified framework can achieve high accuracy and real-time reconstruction for the color image at extremely low subsampling rate.

DSRD: deep sparse representation with learnable dictionary for remotely sensed image denoising

ABSTRACT Patch-based modelling methods (i.e. sparse coding) had proven their great ability in solving the degradation problem caused by additive white Gaussian noise (AWGN) in remotely sensing images. However, these methods usually pursue surprising performance with sacrificing computational efficiency, and the learned dictionary is universal, resulting in a weak model representation flexibility. To address these issues, this paper proposes a deep sparse representation with learnable dictionary (DSRD) scheme, where the major difference from the previous sparse coding methods is that sparse representation coefficients and dictionaries are both deeply learned, acted as two modular parts to be plugged into the unfolded sparse coding model, to speed up for a stable solution as well as yielding competitive performance. The ablation studies illustrate that our DSRD strategy is efficient for a fixed solution while a flexible and powerful capacity in producing pleasing denoising performance. Meanwhile, the comparisons of quantitatively and qualitatively experimental results further demonstrate that it is effective and can generate enjoyable denoising performance that even outperforms that produced by other state-of-the-arts.

Here Today, Gone Tomorrow: A Mixed‐method Study on Headhunter‐intermediated Recruitment Practices

AbstractWhile organizations increasingly rely on headhunters, little research has focused on what kinds of employees are contacted by headhunters and how these contacts are related to turnover. This study uses an exploratory sequential mixed‐method design to examine what kinds of employees contingency headhunters contact and why some of them leave. Using the case of Japan, where foreign subsidiaries use headhunters extensively to recruit host country nationals (HCNs), we first conducted interviews with 131 clients, headhunters and HCNs (candidates). The findings indicated that the contacts made by headhunters were guided by the HCNs’ age, gender, education level, English proficiency, tenure and prior placement by headhunters. The findings also showed that headhunter contacts drive up turnover, and that job alternatives and job embeddedness moderate the relationship between headhunter contacts and turnover. We used these findings and research on the unfolding model and headhunters to develop an abductively derived research model and hypotheses. In another study, we tested these hypotheses by time‐lagged survey data from 456 HCNs in foreign subsidiaries. The survey findings showed that headhunter contacts mediate the relationships between English proficiency, tenure and prior placement by headhunters, and that job alternatives and job embeddedness moderate the positive relationship between headhunter contacts and turnover.

Modified Frantz-Nodvik equation and numerical simulation of a high-power Innoslab picosecond laser amplifier.

A modified Frantz-Nodvik (F-N) equation and a simple one-dimensional unfolded slicing model for numerically simulating high-power Innoslab picosecond amplifier are developed for the first time. The anisotropic stimulated emission cross-section of laser crystal, the influence of the tilted optical path, the spatial overlap of the seed and pump laser, as well as the pump absorption saturation effect are considered. Based on the as-developed model, 4-, 6- and 8-pass schemes high-power Nd:YVO4 Innoslab picosecond amplifiers are designed with output powers of 76.2 W, 81.4 W, and 85.5 W, respectively. The experimental results agree well with that of numerical simulation, indicating that our model is a powerful tool and paves a new way for designing and optimizing high-power Innoslab picosecond laser amplifier.

How Do Employee Attitude Surveys Contribute to Staff Turnover Intentions in a University Setting?

This study aimed to determine the effect of employee attitude surveys on academic staff turnover intentions in chartered universities in Kenya. The specific objectives were; to determine the extent of employee attitude surveys practices among universities in Kenya; assess the level of turnover intentions among academic staff in chartered universities in Kenya, and determine the effect of employee attitude surveys practices on turnover intentions among academic staff in chartered universities in Kenya. The study was anchored on the Universalistic theory and the Unfolding model of voluntary turnover. A positivism research philosophy guided the study, and a descriptive cross-sectional survey design was used. The study obtained primary data from a representative sample of 364 academic staff members drawn from 15 chartered universities in Kenya. The study found that employee attitude surveys have been practized to a low extent and produced correspondingly low staff turnover intentions. Two dimensions of employee attitude surveys significantly negatively affect staff turnover intentions. The study called on future research to apply more robust statistical techniques anchored on mixed methods design for a more comprehensive explanation of the direction of the causal effects of attitude surveys on staff turnover intentions.

Bayesian Approaches for Detecting Differential Item Functioning Using the Generalized Graded Unfolding Model.

Differential item functioning (DIF) analysis is one of the most important applications of item response theory (IRT) in psychological assessment. This study examined the performance of two Bayesian DIF methods, Bayes factor (BF) and deviance information criterion (DIC), with the generalized graded unfolding model (GGUM). The Type I error and power were investigated in a Monte Carlo simulation that manipulated sample size, DIF source, DIF size, DIF location, subpopulation trait distribution, and type of baseline model. We also examined the performance of two likelihood-based methods, the likelihood ratio (LR) test and Akaike information criterion (AIC), using marginal maximum likelihood (MML) estimation for comparison with past DIF research. The results indicated that the proposed BF and DIC methods provided well-controlled Type I error and high power using a free-baseline model implementation, their performance was superior to LR and AIC in terms of Type I error rates when the reference and focal group trait distributions differed. The implications and recommendations for applied research are discussed.