Refinement Process Research Articles

Challenges and promises of self-adaptive simulation models

The notion of (self-)adaptation, according to J. Holland (1975), describes the process of iterative refinement of an entity’s behavior or structure to improve its performance in its environment. We argue that the potential role of (self-)adaptive simulation models has not been sufficiently acknowledged in the past. A focus on adaptive simulation will likely propel methodological advances in modeling and simulation and digital twinning, increasing its impact on solving urgent real-world problems. Therefore, we will review methods for including adaptation as part of or within simulation models and consequently discuss how simulation models themselves may become the subject of adaptation within and across simulation studies. We will identify different motivations for and types of adaptations within simulation studies by analyzing a family of simulation models. The need to automatically conduct various types of adaptations increases with the importance of online adaptations, e.g. as encountered in digital twins. Further methodological developments in unambiguously representing and intelligently processing the various knowledge sources used in simulation studies, domain-specific languages, analysis methods, self-adaptive software, and, last but not least, model learning are needed. Combining the adaptation within and of simulation models, we arrive at the vision of self-adaptive models, which are less subject to adaptation but become the central actors in their adaptation.

Reconstructing compact building models from 3D indoor point cloud with curved surfaces via global energy optimization

The automatic reconstruction of indoor interiors from scanned 3D point clouds has emerged as a substantial challenge. Most indoor reconstruction methods typically focus on planar structures rather than curved surfaces. Certain primitive representation-based approaches reconstruct curved surfaces by fitting and replacing regularity primitives, but they suffer from generating compact, watertight models solely through primitive replacement when occlusion and varying densities are present. This paper addresses indoor curved surface reconstruction by employing a hypothesis and selection strategy under global energy optimization (GEO). Hypothesized cells are generated through a combination of points and supervoxels with adaptive resolutions, thereby enhancing their geometric precision and integrity. Additionally, the optimal cell is selected through iterative refinement and clustering processing steps under GEO, ensuring that the reconstructed model accurately adheres to the target interior structure while maintaining compactness. A series of experiments demonstrate the effectiveness and feasibility of the proposed method.

An improved goal-oriented adaptive finite-element method for 3-D direct current resistivity anisotropic forward modeling using nested tetrahedra

We developed a novel adaptive finite element method (FEM) to address the problem of 3-D direct current (DC) resistivity forward modeling with complex surface topography and arbitrary conductivity anisotropy. The tetrahedra-based FEM and secondary virtual potential algorithm are first used to handle arbitrary complex geo-models. Then, to ensure the accuracy of the simulation solution, an improved goal-oriented adaptive mesh refinement (AMR) algorithm is proposed to realize an optimized mesh density distribution. To avoid the drawback of the traditional goal-oriented AMR algorithm for the DC forward modeling problem, we incorporate a volume-based weighting factor into the posterior error estimation procedure to further optimize the density distribution of the forward modeling grid. In addition, instead of traditional open source mesh generation software, we propose using the longest-edge bisection (LEB) algorithm to perform the mesh refinement process, which can preserve the topological structure between different-level meshes. Finally, the comprehensive test using a two-layered model and two complex 3-D models demonstrate the capability of our newly developed code to obtain highly accurate solutions even on relatively coarse initial grids. By incorporating the volume factor, our novel AMR algorithm achieves a more uniform and reasonable mesh density distribution during these experiments. The LEB refinement technique can generate a series of nested tetrahedral elements and provide fewer tetrahedral elements compared to the traditional Delaunay-based AMR method. The proposed 3-D DC forward modeling method has been implemented into an open source C++ code, which will contribute to the advancement of the 3-D DC resistivity imaging field.

Neural Disparity Refinement.

We propose a framework that combines traditional, hand-crafted algorithms and recent advances in deep learning to obtain high-quality, high-resolution disparity maps from stereo images. By casting the refinement process as a continuous feature sampling strategy, our neural disparity refinement network can estimate an enhanced disparity map at any output resolution. Our solution can process any disparity map produced by classical stereo algorithms, as well as those predicted by modern stereo networks or even different depth-from-images approaches, such as the COLMAP structure-from-motion pipeline. Nonetheless, when deployed in the former configuration, our framework performs at its best in terms of zero-shot generalization from synthetic to real images. Moreover, its continuous formulation allows for easily handling the unbalanced stereo setup very diffused in mobile phones.

Brand Corporateness: Measurement and Symbolic Meaning of an Unfavorable Brand Association

Abstract This research establishes brand corporateness as a novel brand association that most consumers find unfavorable. Exploratory focus group findings first illuminate consumer meaning structures and attitudes around brand corporateness. An inductive grounded theory approach suggests three core (hierarchical, mechanistic, opaque) and three ancillary (ubiquitous, traditional, strategic) dimensions of the construct, informing a subsequent literature review that theoretically confirms this structure. Our theorizing also suggests that greater corporateness diminishes consumers’ self-brand connection because the symbolic meanings of its core dimensions clash with several fundamental human values, a negative effect that is heightened among more politically liberal consumers. The main empirical work then develops and validates a measure of brand corporateness across eleven studies using best practices in scale development. After a rigorous item generation and refinement process, our studies provide converging evidence for the scale’s structural, nomological, discriminant, predictive, and ecological validity.

StableNormal: Reducing Diffusion Variance for Stable and Sharp Normal

This work addresses the challenge of high-quality surface normal estimation from monocular colored inputs (i.e., images and videos), a field which has recently been revolutionized by repurposing diffusion priors. However, previous attempts still struggle with stochastic inference, conflicting with the deterministic nature of the Image2Normal task, and costly ensembling step, which slows down the estimation process. Our method, StableNormal, mitigates the stochasticity of the diffusion process by reducing inference variance, thus producing "Stable-and-Sharp" normal estimates without any additional ensembling process. StableNormal works robustly under challenging imaging conditions, such as extreme lighting, blurring, and low quality. It is also robust against transparent and reflective surfaces, as well as cluttered scenes with numerous objects. Specifically, StableNormal employs a coarse-to-fine strategy, which starts with a one-step normal estimator (YOSO) to derive an initial normal guess, that is relatively coarse but reliable, then followed by a semantic-guided refinement process (SG-DRN) that refines the normals to recover geometric details. The effectiveness of StableNormal is demonstrated through competitive performance in standard datasets such as DIODE-indoor, iBims, ScannetV2 and NYUv2, and also in various downstream tasks, such as surface reconstruction and normal enhancement. These results evidence that StableNormal retains both the "stability" and "sharpness" for accurate normal estimation. StableNormal represents a baby attempt to repurpose diffusion priors for deterministic estimation. To democratize this, code and models have been publicly available in hf.co/Stable-X.

Effects of nano-modification on the microstrfuctural evolution and mechanical properties of AA7075/AA6061 aluminum alloy joints fabrica ted by laser-MIG hybrid welding technique

In this study, a rigorous comparative analysis is conducted to assess the implications of 7075 welding wire and 7075 welding wire fortified with 1.0 wt% of TiC nanoparticles on the microstructural characteristics, distribution patterns of secondary phases, grain refinement, and mechanical properties of AA7075/AA6061 dissimilar aluminum alloy joints produced via laser-MIG hybrid welding. The experimental results show that: The introduction of TiC nanoparticles prompts a transformation in the weld microstructure, transitioning from coarse dendritic morphologies to finer equiaxed dendritic structures. The grain size and the strengthening phases have both undergone a process of refinement. Notably, a marked increase in high-angle grain boundaries (HAGBs) and a concomitant decrease in low-angle grain boundaries (LAGBs) are observed in the nano-modified weld. Nano-modification yields substantial improvements in the mechanical properties of the welded joint. Specifically, the softening width of the AA6061 heat-affected zone (HAZ) is reduced by approximately half (from 8 mm to 4 mm), while the minimum microhardness of the joint has been increased from 34.1 Hv to 62.4 Hv. Moreover, the average ultimate tensile strength (UTS) of the joint has been enhanced from 95.7 MPa to 208.9 MPa, while the average yield strength (YS) has been increased from 57.1 MPa to 155.1 MPa.

Efficient Encoding and Aligning Viewpoints for 6D Pose Estimation of Unseen Industrial Parts

Abstract Estimating the 6D poses of industrial parts is a fundamental task in automated industries. However, the scarcity of industrial part datasets and the effort involved to retrain networks present challenges when estimating poses for unseen parts. Although a few pre-trained networks demonstrate effectiveness on unseen objects, they often struggle to encode correct viewpoint for unseen industrial parts, which have significant geometrical differences compared to the pre-trained objects. Additionally, they overlook the viewpoint non-uniformity that frequently occurs in industrial settings, resulting in significant 3D rotation errors. To address these issues, a novel 6D pose estimator for unseen industrial parts is proposed. First, a Self-to-Inter (S2I) viewpoint encoder is introduced to efficiently generate discriminative descriptors that capture the viewpoint information of the observed image. The S2I viewpoint encoder utilizes an Inter-viewpoint attention module to facilitate prior viewpoint communication and leverages a saliency descriptor selection strategy to boost inference speed. Second, a Viewpoint Alignment module (VAM) is established and integrated with the ICP refiner. The VAM aligns non-uniform viewpoints in an analytical paradigm, leading to enhanced efficiency of the refinement process and more accurate final predictions. Experimental results on the LINEMOD dataset demonstrate competitive performance compared to state-of-the-art methods. Furthermore, the experiments conducted on eight unseen industrial parts validate the exceptional generalizability of our method, highlighting its potential in industrial applications.

ScGraphformer: unveiling cellular heterogeneity and interactions in scRNA-seq data using a scalable graph transformer network

The precise classification of cell types from single-cell RNA sequencing (scRNA-seq) data is pivotal for dissecting cellular heterogeneity in biological research. Traditional graph neural network (GNN) models are constrained by reliance on predefined graphs, limiting the exploration of complex cell-to-cell relationships. We introduce scGraphformer, a transformer-based GNN that transcends these limitations by learning an all-encompassing cell-cell relational network directly from scRNA-seq data. Through an iterative refinement process, scGraphformer constructs a dense graph structure that captures the full spectrum of cellular interactions. This comprehensive approach enables the identification of subtle and previously obscured cellular patterns and relationships. Evaluated on multiple datasets, scGraphformer demonstrates superior performance in cell type identification compared to existing methods and showcases its scalability with large-scale datasets. Our method not only provides enhanced cell type classification ability but also reveals the underlying cell interactions, offering deeper insights into functional cellular relationships. The scGraphformer thus holds the potential to significantly advance the field of single-cell analysis and contribute to a more nuanced understanding of cellular behavior.

Blending is all you need: Data-centric ensemble synthetic data

Deep generative models have gained increasing popularity, particularly in fields such as natural language processing and computer vision. Recently, efforts have been made to extend these advanced algorithms to tabular data. While generative models have shown promising results in creating synthetic data, their high computational demands and the need for careful parameter tuning present significant challenges. This study investigates whether a collective integration of refined synthetic datasets from multiple models can achieve comparable or superior performance to that of a single, large generative model. To this end, we developed a Data-Centric Ensemble Synthetic Data model, leveraging principles of ensemble learning. Our approach involved a data refinement process applied to various synthetic datasets, systematically eliminating noise and ranking, selecting, and combining them to create an augmented, high-quality synthetic dataset. This approach improved both the quantity and quality of the data. Central to this process, we introduced the Ensemble k-Nearest Neighbors with Centroid Displacement (EKCD) algorithm for noise filtering, alongside a density score for ranking and selecting data. Our experiments confirmed the effectiveness of EKCD in removing noisy synthetic samples. Additionally, the ensemble model based on the refined synthetic data substantially enhanced the performance of machine learning models, sometimes even outperforming that of real data.

Prediction of Sulfur Content during Steel Refining Process Based on Machine Learning Methods

The neural network technology combining genetic algorithm is utilized to predict the sulfur content and optimize the desulfurization operation at the end of the refining process. Three types of prediction models are developed to achieve the optimal model. The prediction accuracy can be improved by the application of the deep neural network while the root means square error (RMSE) value of the optimal prediction model and the mean absolute error (MAE) value are less than 5 ppm. Moreover, the proportion of heats with prediction errors less than 5 ppm reaches 82%. Effects of dissolved oxygen contents, initial sulfur contents, carbon contents, and the amount of desulfurizer addition on the desulfurization process are considered. The optimal amount of slag addition with various initial sulfur contents is calculated. With the increase of initial sulfur content in the molten steel, the optimal amount of slag‐modified agent addition increases from about 500–750 kg.

MISNeR: Medical Implicit Shape Neural Representation for Image Volume Visualisation

AbstractThree‐dimensional visualisation of mesh reconstruction of the medical images is commonly used for various clinical applications including pre / post‐surgical planning. Such meshes are conventionally generated by extracting the surface from volumetric segmentation masks. Therefore, they have inherent limitations of staircase artefacts due to their anisotropic voxel dimensions. The time‐consuming process for manual refinement to remove artefacts and/or the isolated regions further adds to these limitations. Methods for directly generating meshes from volumetric data by template deformation are often limited to simple topological structures, and methods that use implicit functions for continuous surfaces, do not achieve the level of mesh reconstruction accuracy when compared to segmentation‐based methods. In this study, we address these limitations by combining the implicit function representation with a multi‐level deep learning architecture. We introduce a novel multi‐level local feature sampling component which leverages the spatial features for the implicit function regression to enhance the segmentation result. We further introduce a shape boundary estimator that accelerates the explicit mesh reconstruction by minimising the number of the signed distance queries during model inference. The result is a multi‐level deep learning network that directly regresses the implicit function from medical image volumes to a continuous surface model, which can be used for mesh reconstruction from arbitrary high volume resolution to minimise staircase artefacts. We evaluated our method using pelvic computed tomography (CT) dataset from two public sources with varying z‐axis resolutions. We show that our method minimised the staircase artefacts while achieving comparable results in surface accuracy when compared to the state‐of‐the‐art segmentation algorithms. Furthermore, our method was 9 times faster in volume reconstruction than comparable implicit shape representation networks.

Evolution of Non-metallic Inclusions in Si-Ca-Ba-Killed Heavy Rail Steel During Secondary Refining Process

Evolution of Non-metallic Inclusions in Si-Ca-Ba-Killed Heavy Rail Steel During Secondary Refining Process

Are customers naturally attracted to brands? A core social motives scale for measuring brand engagement

PurposeTo extend consumer–brand relationship research to the satisfaction of social needs, we introduce a new construct to assess consumer propensity to develop brand relationships based on the satisfaction of social needs through brands and provide a scale to measure this construct.Design/methodology/approachWe implement a two-stage procedure based on the literature and expert judgments to formulate questions based on questionnaire data, and refinement processes and validation analyses are applied to measure the validity and reliability of the scale.FindingsWe developed a scale for measuring consumer propensity for establishing brand relationships that covers five dimensions: belonging, control, understanding, self-enhancement and trust. The scale was validated and found to be reliable.Research limitations/implicationsThis research could benefit from model testing using samples from outside the USA, thus incorporating the role of culture and considering behavioral outputs.Practical implicationsThe identification and operationalization of the propensity to relate to brands are important in market segmentation for more efficiently deepening brand relationships. Additionally, this conceptualization improves our understanding of the elements that contribute to the relationship between brands and consumers: belonging, control, understanding, self-enhancement and trust.Originality/valueAlthough the consumer–brand relationship is recognized in the literature, the satisfaction of social needs via this relationship has been ignored. This paper aims to enhance our understanding of person–brand relationships by proposing a scheme for analyzing the degree of need satisfaction facilitated by this relationship.

Novel Image-Guided Simulator for Transcervical Intralaryngeal Injection Training.

To assess the impact of a novel 3D-printed simulation model with Brainlab Image Guidance on enhancing otolaryngology residents' skills and confidence in performing transcervical intralaryngeal injection (TII) compared with conventional training methods. Utilizing a 3D-printed larynx model derived from computed tomography (CT) scans, this study involved 16 otolaryngology residents divided into two groups for TII training: one with Brainlab Image Guidance (LMIG) and the other without (LM). Pre- and post-training evaluations measured participants' confidence while the Brainlab system measured the accuracy of their needle placements. After training, participants exhibited a significant increase in confidence with an average rise from 1.56 to 2.75 on a 5-point scale. The LMIG group outperformed the LM group in accuracy achieving statistically significant reductions in target distances after training (3.5 mm right, 3.6 mm left). The LMIG also demonstrated a significantly greater increase in procedural confidence over the LM group after training. The TII laryngeal model with Brainlab Image Guidance significantly improves procedural confidence and accuracy among otolaryngology residents, signifying potential advantage over a more conventional training approach. The model's realistic tactile and live instrument positioning feedback augments the process of surgical skill refinement in a controlled, risk-free, simulation environment. NA Laryngoscope, 2024.

Diffusion process with structural changes for subspace clustering

Spectral clustering-based methods have gained significant popularity in subspace clustering due to their ability to capture the underlying data structure effectively. Standard spectral clustering focuses on only pairwise relationships between data points, neglecting interactions among high-order neighboring points. Integrating the diffusion process can address this limitation by leveraging a Markov random walk. However, ensuring that diffusion methods capture sufficient information while maintaining stability against noise remains challenging. In this paper, we propose the Diffusion Process with Structural Changes (DPSC) method, a novel affinity learning framework that enhances the robustness of the diffusion process. Our approach broadens the scope of nearest neighbors and leverages the dropout idea to generate random transition matrices. Furthermore, inspired by the structural changes model, we use two transition matrices to optimize the iteration rule. The resulting affinity matrix undergoes self-supervised learning and is subsequently integrated back into the diffusion process for refinement. Notably, the convergence of the proposed DPSC is theoretically proven. Extensive experiments on benchmark datasets demonstrate that the proposed method outperforms existing subspace clustering methods. The code of our proposed DPSC is available at https://github.com/zhudafa/DPSC.

Addressing ethical issues in healthcare artificial intelligence using a lifecycle-informed process.

Artificial intelligence (AI) proceeds through an iterative and evaluative process of development, use, and refinement which may be characterized as a lifecycle. Within this context, stakeholders can vary in their interests and perceptions of the ethical issues associated with this rapidly evolving technology in ways that can fail to identify and avert adverse outcomes. Identifying issues throughout the AI lifecycle in a systematic manner can facilitate better-informed ethical deliberation. We analyzed existing lifecycles from within the current literature for ethical issues of AI in healthcare to identify themes, which we relied upon to create a lifecycle that consolidates these themes into a more comprehensive lifecycle. We then considered the potential benefits and harms of AI through this lifecycle to identify ethical questions that can arise at each step and to identify where conflicts and errors could arise in ethical analysis. We illustrated the approach in 3 case studies that highlight how different ethical dilemmas arise at different points in the lifecycle. Through case studies, we show how a systematic lifecycle-informed approach to the ethical analysis of AI enables mapping of the effects of AI onto different steps to guide deliberations on benefits and harms. The lifecycle-informed approach has broad applicability to different stakeholders and can facilitate communication on ethical issues for patients, healthcare professionals, research participants, and other stakeholders.

Lifing Assessment of Gas Turbine Blade Root Affected by Out-of-Tolerances.

Current and future heavy-duty gas turbines (GTs) are being developed as an alternative or support to renewable energy sources (RESs). Therefore, GTs are subjected to several instances of being switched on and off; thus, the material fatigue limit can be reached in a short time. In such a scenario, possible out-of-tolerances (OoTs) in critical components must be considered. In this paper, OoTs related to critical parameters in the attachment geometry of a rotor blade are considered to estimate their impact on component life through a 2D finite element (FE) analysis. First, a mesh refinement is performed to obtain mesh-independent results; second, OoT geometries are simulated to determine stresses and strains at the blade attachment and disc groove. The mesh refinement process is critical to ensure model accuracy for both nominal and OoT geometries. The results show that OoTs can lead to important reductions in the life of the intended component, both on the blade and disc sides. These results could be useful in updating the maintenance plan for components and could be used for future insights, with further work extending the study to 3D geometry, for example, and evaluating the effect of other geometries.

Refinement and Validation of the Team Effectiveness Scale for Nursing Units.

Understanding that the complexity and dynamic nature of the nursing care setting creates diverse conditions for teamwork is important when developing tools to measure nursing unit team effectiveness. The Team Effectiveness Scale for Nursing Units (TES-NU), based on the Integrated Team Effectiveness Model, was developed without confirmatory factor analysis and only tested on one nursing organization. It needs further research to prove its validity and reliability. This study aims to refine and validate the TES-NU in various nursing organizations. We designed this methodological study to refine the TES-NU by establishing its validity and reliability. The study included 330 clinical nurses from six general hospitals in South Korea, selected via convenience sampling. The TES-NU's refinement process includes item analysis, exploratory factor analysis, confirmatory factor analysis, item analysis, and convergent validity. The KMO of 22 preliminary items was 0.89, the cumulative variance of the five factors was 67.58%, and the commonality was >0.40. Confirmatory factor analysis indicated the revised model fit well with better indices: CMIN/DF = 1.687, CFI = 0.936, TLI = 0.924, RMSEA = 0.059, and SRMR = 0.057. We simplified the refined scale to 22 items in 5 subdomains: "head nurses leadership", "job satisfaction", "cohesion", "work performance", and "nurses competence". Convergent validity (r = 0.69, p < 0.001) and reliability (Cronbach's alpha = 0.92) were validated for the revised TES-NU. A refined TES-NU has tested their validity and reliability. Nursing managers can use this tool to manage the performance of individual nurses as well as nursing units, which will contribute to improving the work performance of the nursing organization.

Proposed Nutrition Competencies for Medical Students and Physician Trainees: A Consensus Statement.

In 2022, the US House of Representatives passed a bipartisan resolution (House of Representatives Resolution 1118 at the 117th Congress [2021-2022]) calling for meaningful nutrition education for medical trainees. This was prompted by increasing health care spending attributed to the growing prevalence of nutrition-related diseases and the substantial federal funding via Medicare that supports graduate medical education. In March 2023, medical education professional organizations agreed to identify nutrition competencies for medical education. To recommend nutrition competencies for inclusion in medical education to improve patient and population health. The research team conducted a rapid literature review to identify existing nutrition-related competencies published between July 2013 and July 2023. Additional competencies were identified from learning objectives in selected nutrition, culinary medicine, and teaching kitchen curricula; dietetic core competencies; and research team-generated de novo competencies. An expert panel of 22 nutrition subject matter experts and 15 residency program directors participated in a modified Delphi process and completed 4 rounds of voting to reach consensus on recommended nutrition competencies, the level of medical education at which they should be included, and recommendations for monitoring implementation and evaluation of these competencies. A total of 15 articles met inclusion criteria for competency extraction and yielded 187 competencies. Through review of gray literature and other sources, researchers identified 167 additional competencies for a total of 354 competencies. These competencies were compiled and refined prior to voting. After 4 rounds of voting, 36 competencies were identified for recommendation: 30 at both undergraduate and graduate levels, 2 at the undergraduate level only, and 4 at the graduate level only. Competencies fell into the following nutrition-related themes: foundational nutrition knowledge, assessment and diagnosis, communication skills, public health, collaborative support and treatment for specific conditions, and indications for referral. A total of 36 panelists (97%) recommended nutrition competencies be assessed as part of licensing and board certification examinations. These competencies represent a US-based effort to use a modified Delphi process to establish consensus on nutrition competencies for medical students and physician trainees. These competencies will require an iterative process of institutional prioritization, refinement, and inclusion in current and future educational curricula as well as licensure and certification examinations.