Full Model Research Articles

Overcoming the Barrier of Incompleteness: A Hyperspectral Image Classification Full Model.

Deep learning-based methods have shown promising outcomes in many fields. However, the performance gain is always limited to a large extent in classifying hyperspectral image (HSI). We discover that the reason behind this phenomenon lies in the incomplete classification of HSI, i.e., existing works only focus on a certain stage that contributes to the classification, while ignoring other equally or even more significant phases. To address the above issue, we creatively put forward three elements needed for complete classification: the extensive exploration of available features, adequate reuse of representative features, and differential fusion of multidomain features. To the best of our knowledge, these three elements are being established for the first time, providing a fresh perspective on designing HSI-tailored models. On this basis, an HSI classification full model (HSIC-FM) is proposed to overcome the barrier of incompleteness. Specifically, a recurrent transformer corresponding to Element 1 is presented to comprehensively extract short-term details and long-term semantics for local-to-global geographical representation. Afterward, a feature reuse strategy matching Element 2 is designed to sufficiently recycle valuable information aimed at refined classification using few annotations. Eventually, a discriminant optimization is formulized in accordance with Element 3 to distinctly integrate multidomain features for the purpose of constraining the contribution of different domains. Numerous experiments on four datasets at small-, medium-, and large-scale demonstrate that the proposed method outperforms the state-of-the-art (SOTA) methods, such as convolutional neural network (CNN)-, fully convolutional network (FCN)-, recurrent neural network (RNN)-, graph convolutional network (GCN)-, and transformer-based models (e.g., accuracy improvement of more than 9% with only five training samples per class). The code will be available soon at https://github.com/jqyang22/ HSIC-FM.

Developing screening tools to estimate the risk of diabetic kidney disease in patients with type 2 diabetes mellitus.

Diabetic kidney disease (DKD) is an important microvascular complication of diabetes mellitus (DM). This study aimed to develop predictive nomograms to estimate the risk of DKD in patients with type 2 diabetes mellitus (T2DM). The medical records of patients with T2DM in our hospital from March 2022 to March 2023 were retrospectively reviewed. The enrolled patients were randomly selected for training and validation sets in a 7:3 ratio. The models for predicting risk of DKD were virtualized by the nomograms using logistic regression analysis. Among the enrolled 597 patients, 418 were assigned to the training set, while 179 were assigned to the validation set. Using the predictors included glycated hemoglobin A1c (HbA1c), high density lipoprotein cholesterol (HDL-C), presence of diabetic retinopathy (DR) and duration of diabetes (DD), we constructed a full model (model 1) for predicting DKD. And using the laboratory indexes of HbA1c, HDL-C, and cystatin C (Cys-C), we developed a laboratory-based model (model 2). The C-indexes were 0.897 for model 1 and 0.867 for model 2, respectively. The calibration curves demonstrated a good agreement between prediction and observation in the two models. The decision curve analysis (DCA) curves showed that the two models achieved a net benefit across all threshold probabilities. We successfully constructed two prediction models to evaluate the risk of DKD in patients with T2DM. The two models exhibited good predictive performance and could be recommended for DKD screening and early detection.

Byzantine Agreement with Optimal Resilience via Statistical Fraud Detection

Since the mid-1980s it has been known that Byzantine Agreement can be solved with probability 1 asynchronously, even against an omniscient, computationally unbounded adversary that can adaptively corrupt up to f < n/3 parties. Moreover, the problem is insoluble with f ≥ n/3 corruptions. However, Bracha’s [ 13 ] 1984 protocol (see also Ben-Or [ 8 ]) achieved f < n/3 resilience at the cost of exponential expected latency 2 Θ ( n ) , a bound that has never been improved in this model with f = ⌊ (n-1)/3 ⌋ corruptions. In this article, we prove that Byzantine Agreement in the asynchronous, full information model can be solved with probability 1 against an adaptive adversary that can corrupt f < n/3 parties, while incurring only polynomial latency with high probability . Our protocol follows an earlier polynomial latency protocol of King and Saia [ 33 , 34 ], which had suboptimal resilience, namely f ≈ n /10 9 [ 33 , 34 ]. Resilience f = (n-1)/3 is uniquely difficult, as this is the point at which the influence of the Byzantine and honest players are of roughly equal strength. The core technical problem we solve is to design a collective coin-flipping protocol that eventually lets us flip a coin with an unambiguous outcome. In the beginning, the influence of the Byzantine players is too powerful to overcome, and they can essentially fix the coin’s behavior at will. We guarantee that after just a polynomial number of executions of the coin-flipping protocol, either (a) the Byzantine players fail to fix the behavior of the coin (thereby ending the game) or (b) we can “blacklist” players such that the blacklisting rate for Byzantine players is at least as large as the blacklisting rate for good players. The blacklisting criterion is based on a simple statistical test of fraud detection .

Adverse childhood event subtypes and routine diabetes care utilization among U.S. adults with Type 2 diabetes.

Adverse childhood events (ACEs) have significant impacts on adulthood health, including greater risk of Type 2 diabetes (T2D). While there is a known connection to increased use of outpatient and emergency healthcare services, the potential role of ACEs in routine diabetes care utilization remains unclear. This study uses ACE subtypes to explain pathways to routine diabetes care utilization among adults with T2D. Cross-sectional data were obtained from the 2019 Behavioral Risk Factor Surveillance System, a survey of a representative sample of U.S. adults. Eligible participants resided in a state that completed both the Adverse Childhood Experiences (ACEs) and Diabetes modules and were diagnosed with non-gestational diabetes (N = 9,904), of whom the majority were presumed to have T2D. Confirmatory factor analysis was used to test the measurement model for three ACE factors: household dysfunction, physical and emotional abuse, and sexual abuse. Structural equation modeling was used to relate factors to four routine diabetes care utilization outcomes. Factor loadings were strong, and fit indices indicated good measurement model and full structural model fits. In the full structural equation model, household dysfunction was associated with decreased likelihood of meeting frequency recommendations for A1C testing (β = -0.22, p < .05) and foot exams (β = -0.29, p < .01). Physical and emotional abuse were associated with greater likelihood of meeting A1C testing frequency recommendations (β = 0.21, p < .05). Specific types of ACEs may differentially relate to routine diabetes care utilization in adulthood. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Predicting Soluble Solids Concentration of ‘Geneva 3’ Kiwiberries Using Near Infrared Spectroscopy

Near infrared (NIR) spectroscopy can be applied to nondestructively assess soluble solids concentration (SSC) of ripening, physiologically mature ‘Geneva 3’ kiwiberries (Actinidia arguta). Spectrographic signatures were captured using a handheld NIR produce quality meter to build predictive models of internal fruit quality for ‘Geneva 3’ kiwiberries that had been held under cold storage (CS) conditions (0 to 1 °C, &gt;90% relative humidity) as well as those not subjected to CS. The CS model, constructed using scans of 133 berries following 4 to 6 weeks in CS, predicts SSC using NIR wavelengths in the range of 729 to 975 nm. A total of 507 berries fresh from the vine were used to construct a predictive model for SSC of non-CS fruit using the same wavelength range. In each case, model predictive performance was investigated using split-half cross-validation, resulting in mean absolute error (MAE) values of 1.2% and 0.8% SSC for the CS and non-CS model, respectively. Each full model was then used to predict SSC of kiwiberries subjected to the alternative CS condition. The non-CS model maintained a low MAE (1.6% SSC) when applied to CS fruit, but the MAE of the CS model applied to non-CS fruit rose considerably (4.5% SSC). The performance of a combined model was tested against both CS and non-CS models, and a benefit to using tailored, CS-specific models was found, particularly in light of cross-seasonal results. As it has proven in many crops, NIR spectroscopy appears to be a promising tool for nondestructively assessing SSC in ‘Geneva 3’ kiwiberry fruit, with accuracy being enhanced by training models specific to postharvest regimes and/or defined ranges of SSC.

Causal Incremental Graph Convolution for Recommender System Retraining.

The real-world recommender system needs to be regularly retrained to keep with the new data. In this work, we consider how to efficiently retrain graph convolution network (GCN)-based recommender models that are state-of-the-art techniques for the collaborative recommendation. To pursue high efficiency, we set the target as using only new data for model updating, meanwhile not sacrificing the recommendation accuracy compared with full model retraining. This is nontrivial to achieve since the interaction data participates in both the graph structure for model construction and the loss function for model learning, whereas the old graph structure is not allowed to use in model updating. Toward the goal, we propose a causal incremental graph convolution (IGC) approach, which consists of two new operators named IGC and colliding effect distillation (CED) to estimate the output of full graph convolution. In particular, we devise simple and effective modules for IGC to ingeniously combine the old representations and the incremental graph and effectively fuse the long- and short-term preference signals. CED aims to avoid the out-of-date issue of inactive nodes that are not in the incremental graph, which connects the new data with inactive nodes through causal inference. In particular, CED estimates the causal effect of new data on the representation of inactive nodes through the control of their collider. Extensive experiments on three real-world datasets demonstrate both accuracy gains and significant speed-ups over the existing retraining mechanism.

A Conditional Process Analysis of Microaffirmations, Microaggressions, and Mental Health Among Thai Sexual Minorities

ABSTRACT Sexual orientation (SO) microaggressions contribute to mental health issues among sexual minorities. Microaffirmations may mitigate these effects, with internalized heterosexism and SO concealment proposed as mediators. A community sample of 307 Thai sexual minorities completed measures assessing SO microaggressions, microaffirmations, internalized heterosexism, SO concealment, and mental health concerns. Serial mediation analysis using Hayes’ PROCESS macro model 6 tested indirect effects through proposed mediators. Conditional process analysis using PROCESS model 85 examined the moderating role of microaffirmations. These models tested hypothesized moderated serial mediation relationships among study variables. SO microaggressions had a total effect on mental health concerns, directly and indirectly through increased SO concealment. The internalized heterosexism → SO concealment sequence mediated this relationship. Microaffirmations moderated the direct microaggressions-mental health link, reducing this association at higher levels of microaffirmations. The full model accounted for 31.6% of the variance (R2 = 0.316) in mental health concerns. The Johnson-Neyman technique identified 0.613 as the microaffirmations value above which the effect of SO microaggressions on mental health was no longer significant. Findings elucidate mechanisms linking SO microaggressions to mental health issues and microaffirmations’ protective role among Thai sexual minorities. These results could inform efforts to mitigate minority stress impacts.

A Graph-Based Technique for the Automated Control-Oriented Modeling of District Heating Networks

Abstract Advanced control strategies for delivering heat to users in a district heating network have the potential to improve performance and reduce wasted energy. To enable the design of such controllers, this paper proposes an automated plant modeling framework that captures the relevant system dynamics, while being adaptable to any network configuration. Starting from the network topology and system parameters, the developed algorithm generates a state-space model of the system, relying on a graph-based technique to facilitate the combination of component models into a full network model. The accuracy of the approach is validated against experimental data collected from a laboratory-scale district heating network. The verification shows an acceptable average normalized root-mean-square error of 0.39 in the mass flow rates delivered to the buildings, and 0.15 in the network return temperature. Furthermore, the ability of the proposed modeling technique to rapidly generate models characterizing different network configurations is demonstrated through its application to topology optimization. The optimal design, obtained via a branch and bound algorithm, reduces network heat losses by 15% as compared to the conventional length-minimized topology.

Interactive simulation and visualization of point spread functions in single molecule imaging

The point spread function (PSF) is fundamental to any type of microscopy, most importantly so for single-molecule localization techniques, where the exact PSF shape is crucial for precise molecule localization at the nanoscale. Optical aberrations and fixed fluorophore dipoles often result in non-isotropic and distorted PSFs, impairing and biasing conventional fitting approaches. Further, PSF shapes are deliberately modified in PSF engineering approaches for providing improved sensitivity, e.g., for 3D localization or determination of dipole orientation. As this can lead to highly complex PSF shapes, a tool for visualizing expected PSFs would facilitate the interpretation of obtained data and the design of experimental approaches. To this end, we introduce a comprehensive and accessible computer application that allows for the simulation of realistic PSFs based on the full vectorial PSF model. Our tool incorporates a wide range of microscope and fluorophore parameters, including orientationally constrained fluorophores, as well as custom aberrations, transmission and phase masks, thus enabling an accurate representation of various imaging conditions. An additional feature is the simulation of crowded molecular environments with overlapping PSFs. Further, our app directly provides the Cramér–Rao bound for assessing the best achievable localization precision under given conditions. Finally, our software allows for the fitting of custom aberrations directly from experimental data, as well as the generation of a large dataset with randomized simulation parameters, effectively bridging the gap between simulated and experimental scenarios, and enhancing experimental design and result validation.

Multi-criteria structural optimization of a hood inner structure using machine learning technique

This article presents an efficient multi-criteria structural optimization process based on machine learning for shape and size optimization of the hood inner. For this purpose, a low dimensional parameterization based on two-dimensional mapping and dimensionality reduction is used to decrease the number of design variables. There are many design parameters to be considered for designing this structure. In this study, the effect of three important criteria consisting of head injury, eigenfrequency, and static load were analyzed. A full vehicle finite element model is prepared for a newly developed product because head injury analysis requires a full vehicle model due to contact with the inner components of the engine compartment. Then, an experimental study is performed based on the real test scenario for model verification. A new rating method for analyzing the head injury criteria evaluated in different points is proposed and different machine learning techniques are compared in the approximation of the objective function. Finally, the problem is solved as both single and multi-objective optimization problems. Results present considerable improvement in accuracy and efficiency in the identified structure while the method has a low computational cost and the implementation is not difficult.

Clinical research imperatives: principles and priorities from the perspective of Allied Health executives and managers.

Objective It is widely understood that a key means of improving health systems and patient outcomes is through research - accessing, understanding, generating and applying research evidence-based practice. To promote more targeted and strategic research in Allied Health practice, this study sought to establish the principles, areas and priorities for clinical research as perceived by Allied Health leaders in the South Australian public health system. Methods The study used a mixed-methods design (full, sequential and equal model). Participants were recruited from theSouth Australian Department of Health and Wellbeing employment lists for Allied Health senior leaders. Consenting participants attended face-to-face focus groups; after an overview presentation, they discussed the principles for Allied Health research, followed by areas and priorities for research. Summaries of the responses were themed descriptively and circulated electronically so participants could confirm the research areas and ascribe priority ratings, clinical populations and services. Results A total of 28 people attended the stakeholder forum (5 online); 20 responded to the second-round electronic summary. Nine principles of research action were agreed. Fourteen research areas were identified and prioritised. There was a relatively consistent prioritisation of measuring Allied Health value, Allied Health workforce, hospital avoidance and closing the gap for Aboriginal health outcomes - whether the individuals were thinking about their own service or the state as a whole. Conclusions Allied Health leadership identified key principles and priorities for research to improve service delivery and patient outcomes. These priorities should generate further discussion and interest for novice and experienced researchers and leaders and can be used to inform granting and project plans.

First Advanced Bilayer Scaffolds for Tailored Skin Tissue Engineering Produced via Electrospinning and Melt Electrowriting

AbstractIn vitro skin models are validated methods for screening cosmetics and pharmaceuticals, but still have limitations. The bilayer poly(ε‐caprolactone) scaffold/membrane model described here overcomes some of these deficits by integrating a solution electrospun (SES) membrane at the dermoepidermal interface and a melt electrowritten (MEW) scaffold that provides an optimal open‐pore environment for the dermis. To the knowledge, this scaffold/membrane model is the only one capable of creating a properly differentiated, full thickness skin model with neosynthesized extracellular matrix (ECM) in only 18 days. Both the wavy and straight fiber scaffold designs create a well‐organized dermis, but dermal collagen organization differs between designs. Adding cells and vitamin C to the scaffolds improves the mechanical properties to more closely mimic native human skin. These findings establish bicomponent scaffolds as a promising advancement for rapidly creating different skin models with varied properties. The versatility and adaptability of the described model can be used for studying how the biological and physical microenvironment impact skin, and testing dermo‐cosmetics and pharmaceutical treatments on different ages of skin. Furthermore, it can be an excellent new tool for studying wound healing and development into its use as a graft or wound dressing is ongoing.

Indispensability of cross-shell contributions in neutron resonance spacing

Spin- and parity-dependent nuclear level densities (NLDs) are obtained for a configuration interaction shell model using a numerically efficient spectral distribution method. The calculations are performed for 24Na and 25,26,27Mg nuclei using full sd-pf model space that incorporates the cross-shell excitations from the sd to the pf-shell. The obtained NLDs are then employed to determine the s-wave neutron resonance spacing (D0), which is one of the crucial inputs for the predictions of astrophysical reaction rates. Although the considered nuclei are not neutron-rich, the contributions from cross-shell excitations to the pf-shell are indispensable for explaining the experimental data for D0, which otherwise are significantly overestimated.

Development of an Abbreviated Model for Predicting Functional Movement Screen Score Within Tactical Populations.

Thompson, MB, Johnson, QR, Lindsay, KG, and Dawes, JJ. Development of an abbreviated model for predicting functional movement screen score within tactical populations. J Strength Cond Res 38(3): 607-611, 2024-The Functional Movement Screen (FMS) is a tool commonly used to identify compensations when performing 7 specific movement patterns. Timely administration of the full FMS is largely dependent on the practitioner's familiarity and experience with the screening battery. When working in populations that are time-poor (i.e., tactical professionals), administration of the full movement pattern battery is not always feasible. The purpose of this study was to determine which, if any, combination of movement patterns that comprise the FMS could be used to predict total score on this screen among first responders. Functional Movement Screen scores for 99 male subjects (mean ± SD; age: 37.55 ± 9.83 years; height: 180.38 ± 6.59 cm; and body mass: 97.87 ± 15.32 kg) and 9 female subjects (age: 33.22 ± 3.99 years; height: 172.11 ± 8.19 cm; and body mass: 83.99 ± 14.09 kg) from a single law enforcement and fire agency were used for this analysis. Subjects performed the full FMS, which consisted of the following movement patterns: deep squat (DS), hurdle step (HS), in-line lunge (ILL), shoulder mobility (SM), active straight leg raise (ASLR), trunk stability push-up (PU), and rotary stability (RS). A stepwise regression was used to determine the best predictors of the FMS 7-pattern model based on the model's R2. Cronbach's alpha and Guttman's lambda-2 were used to determine the reliability of the proposed models. The regression indicated that a 4-pattern model consisting of DS, ILL, SM, and PU was sufficient to predict approximately 84% of the full model (adjusted R2 = 0.84, p ≤ 0.001). This 4-pattern model was shown to be reliable with the 7-pattern model (α = 0.93, λ = 0.93). Using this modified version of the FMS may allow practitioners working in tactical populations a time-efficient method of identifying dysfunctional movement and determine whether the full 7-pattern model of the FMS should be considered.

Equivalent linearization of hyperelastic rubber rings in dynamic vibration absorber boring bar and simulation of its dynamic properties

A slight perturbation in high-speed precision cutting could cause violent vibration of boring bars, resulting in unexpected wear and failure of the structures. In this paper, the dynamical characteristic of a boring bar with a dynamic vibration absorber (DVA) is investigated. Considering that the nonlinear characteristics of the rubber ring play a significant role in improving the capacity of DVA, we proposed an equivalent linearization method for modeling hyperelastic rubber rings based on the experimental data of uniaxial tensile test. The relationship between the stiffness of rubber rings and the precompression parameter δ is established for two types of rubber. The equivalent linearization model shows good accuracy and calculation efficiency while the mass of the DVA is confined to a narrow space and the assumption of small strain is inherently satisfied. Numerical results show that the time cost of the equivalent linearization model is only 5% of the full hyperelastic model. Furthermore, the effect of different spring stiffnesses and damping on the dynamic characteristics of the boring bar is compared by using the equivalent linearization model, and the vibration peak of the DVA boring bar is found to be reduced about 45%. Our findings establish a simplified modeling scheme for choosing rubber rings in design of dynamic vibration absorbers with acceptable accuracy and little time cost, which could also give some reference for modeling of metal rubber and gel-like materials.

On efficient modelling of radical production in cavitation assisted reactors

Process intensification by cavitation is gaining widespread attention due to the benefits that the intense bubble collapse conditions can provide, yet, several knowledge gaps exist in the modelling of such systems. This work studies the numerical prediction of single bubble dynamics and the various approaches that can be employed to estimate the changes in the chemical composition of cavitating bubbles. Specific emphasis is placed on the prediction of the radical production rates during bubble collapse and the computational performance, with the aim of coupling the single bubble dynamics to flow models for reactor hydrodynamics. The results reveal that the choice of chemical reaction approach has virtually no effect on the bubble dynamics, whereas the predicted radical production rates can differ substantially. It is found that evaluating the radical production only on temperature peaks, an approach commonly followed in literature, may result in the most erroneous estimations (on average 12.8 times larger than those of the full kinetic model), while a simplified kinetic model yields more accurate predictions (2.3 times larger) at the expense of increased computational times. Continuous evaluation of the bubble content by assuming equilibrium when the bubble temperature is above a certain threshold (≈1500K) is shown to be capable of predicting total radical production values close to those estimated by solving the kinetics of a detailed reaction model (19.8% difference), as well as requiring only 22.2% more computational costs compared to simulations without chemical reaction modelling. Such an equilibrium approach is therefore recommended for future studies aiming to couple flow simulations with single bubble dynamics to accurately predict radical production rates in cavitation devices, involving numerous bubbles following different flow trajectories. Furthermore, an algebraic expression that successfully approximates the full kinetic simulation results is proposed as a function of the initial nucleus size and the time integral of the liquid pressure when it is under vapor pressure. Such a model can be applied in modelling efforts that do not require local instantaneous radical concentrations, and paves the way for efficient closure modelling of radical production in CFD simulations of hydrodynamic reactors.

Bayesian Estimation of Topological Features of Persistence Diagrams

Persistent homology is a common technique in topological data analysis providing geometrical and topological information about the sample space. All this information, known as topological features, is summarized in persistence diagrams, and the main interest is in identifying the most persisting ones since they correspond to the Betti number values. Given the randomness inherent in the sampling process, and the complex structure of the space where persistence diagrams take values, estimation of Betti numbers is not straightforward. The approach followed in this work makes use of features’ lifetimes and provides a full Bayesian clustering model, based on random partitions, in order to estimate Betti numbers. A simulation study is also presented.

Equivalent Reduced DC Network Models With Nonlinear Load Functions: A Data-Driven Approach

Nonconvex AC power flow models feature larger computation burden and lack optimality guarantees to be considered in industry expansion and operational planning activities. Additionally, the consideration of full network models, in general, leads to undesirable levels of computational burden, even under the standard DC approximations. Notwithstanding, current reduction methods fail in accurately reproducing real system responses for a wide range of operating points. In this context, we propose a novel data-driven network reduction method to generate equivalent reduced DC power flow models with nonlinear load functions. Our method aims at providing a high-quality representation of the internal system by minimizing the mismatch between the response of the equivalent DC model and the complete AC power flow (or real measurements) under multiple operating-point data. Thus, the approach can be interpreted as a physics-informed machine-learning method as it trains (or estimates) some parameters of the reduced DC power-flow model to fit pre-calculated data. The method determines: 1) a reduced external equivalent network model (topology and parameters), 2) the optimal allocation of external power injections to the boundary buses, and 3) the coefficients determining artificial nonlinear load functions of the operating-point data. Out-of-sample evaluation tests corroborate the performance of our model against modified Ward reduction benchmark models.

The function of power: A herstorical model of power, trauma, and policing African Americans.

Transmitting trauma narratives is intergenerational by nature. Few studies have taken a qualitative approach to analyzing the most pervasive trauma of the United States: the chattel slavery of Black people. Examining the lived experiences of the formerly enslaved, through their childhood and adult narrative memories of personal and second-hand interactions with White authority figures, is critical to the recognition of today's ongoing impact of policing, generational trauma, and mental health in the African American community. Using interviews archived in the Library of Congress from women (N = 19) who were identified as being members of the last living generation of formerly enslaved African Americans, researchers used a feminist-forward grounded theory methodology to understand the following: (a) What are the historical relationships between African Americans and White authority figures? (b) What are the memories associated with the interactions between African Americans and White authority figures? (c) How did these interaction processes come to be? and (d) How did these memories get passed on? The analysis of childhood memories and interactions between enslavers and African Americans close to home (i.e., on the plantation) produced six themes: (a) enmeshment with enslaver, (b) enslaver as good, (c) caretaking by enslaver, (d) enslaver control, (e) violent control by policing figures, and (f) following orders. Theoretical coding led to the formation of a full model of the function of power within the herstorical policing of African Americans close to home. The themes uncovered highlight the potentially traumatic violence and control that characterized the environment in which enslaved children lived and the systems through which White power was maintained. The herstorical analysis and results confirm the ways police violence has sought to control and harm African Americans for over 400 years and underscores the role that police violence has played in the perpetuation of intergenerational trauma and the maintenance of White power. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Decoupling Shrinkage and Selection in Gaussian Linear Factor Analysis

Factor analysis is a popular method for modeling dependence in multivariate data. However, determining the number of factors and obtaining a sparse orientation of the loadings are still major challenges. In this paper, we propose a decision-theoretic approach that brings to light the relationship between model fit, factor dimension, and sparse loadings. This relation is done through a summary of the information contained in the multivariate posterior. A two-step strategy is used in our method. First, given the posterior samples from the Bayesian factor analysis model, a series of point estimates with a decreasing number of factors and different levels of sparsity are recovered by minimizing an expected penalized loss function. Second, the degradation in model fit between the posterior of the full model and the recovered estimates is displayed in a summary. In this step, a criterion is proposed for selecting the factor model with the best trade-off between fit, sparseness, and factor dimension. The findings are illustrated through a simulation study and an application to personality data. We used different prior choices to show the flexibility of the proposed method.