Geometric Loss Research Articles

Progressive evolution of the viscous dissipation mechanism from the macroscale to the nanoscale

Recent studies of viscous dissipation mechanisms in impacting droplets have revealed distinct behaviours between the macroscale and nanoscale. However, the transition of these mechanisms from the macroscale to the nanoscale remains unexplored due to limited research at the microscale. This work addresses the gap using the many-body dissipative particle dynamics (MDPD) method. While the MDPD method omits specific atomic details, it retains crucial mesoscopic effects, making it suitable for investigating the impact dynamics at the microscale. Through the analysis of velocity contours within impacting droplets, the research identifies three primary contributors to viscous dissipation during spreading: boundary-layer viscous dissipation from shear flow; rim geometric head loss; and bulk viscous dissipation caused by droplet deformation. This prompts a re-evaluation of viscous dissipation mechanisms at both the macroscale and nanoscale. It reveals that the same three kinds of dissipation are present across all scales, differing only in their relative intensities at each scale. A model of the maximum spreading factor (βmax) incorporating all forms of viscous dissipation without adjustable parameters is developed to substantiate this insight. This model is validated against three distinct datasets representing the macroscale, microscale and nanoscale, encompassing a broad spectrum of Weber numbers, Ohnesorge numbers and contact angles. The satisfactory agreement between the model predictions and the data signifies a breakthrough in establishing a universal βmax model applicable across all scales. This model demonstrates the consistent nature of viscous dissipation mechanisms across different scales and underscores the importance of integrating microscale behaviours to understand macroscale and nanoscale phenomena.

Signal detection

Acoustic emission (AE) occurring in concrete fracture process is a combination of concrete fracture at local material level and the propagation of induced elastic waves at structural level. Simulation of the fracture-induced AE in concrete needs a model that is able to address the fracture behavior and the wave propagation at the same time. However, to date, no available modelling tools can simulate the whole fracture-induced AE process. Among others, the lattice models have the potential to achieve the simulation of the whole fracture-induced AE process. However, the lattice models are currently restricted to statistical analysis of AE events in fracture process. This paper investigates the feasibility and limitation of lattice model for the simulation of elastic wave propagation process. Several simulation cases are performed considering the propagation of both plane waves and point-source waves. The simulation results, in terms of wave velocity and geometric spreading loss of wave amplitudes, agree well with the analytical solutions. This study represents the first step for the simulation of complete fracture-induced AE waves in concrete.

Second-Line Uterotonics for Uterine Atony: A Randomized Controlled Trial.

To evaluate the comparative efficacy of two of the most commonly used second-line uterotonics-methylergonovine maleate and carboprost tromethamine. We conducted a double-blind randomized trial at two large academic perinatal centers in patients undergoing nonemergency cesarean delivery with uterine atony refractory to oxytocin, as diagnosed by the operating obstetrician. The intervention included administration of a single dose of intramuscular methylergonovine or carboprost intraoperatively at diagnosis. The primary outcome, uterine tone on a 0-10 numeric rating scale 10 minutes after study drug administration, was rated by operating obstetricians blinded to the drug administered. Secondary outcomes included uterine tone score at 5 minutes, administration of additional uterotonic agents, other interventions for uterine atony or hemorrhage, quantitative blood loss, urine output, postpartum change in serum hematocrit, transfusion, length of hospital stay, adverse drug or transfusion reactions, and postpartum hemorrhage complications. A sample size of 50 participants per group was planned to detect a 1-point difference (with estimated within-group SD of 1.5) in the mean primary outcome with 80% power at a two-sided α level of 0.05 while accounting for potential protocol violations. A total of 1,040 participants were enrolled, with 100 randomized to receive one of the study interventions. Mean±SD 10-minute uterine tone scores were 7.3±1.7 after methylergonovine and 7.6±2.1 after carboprost, with an adjusted difference in means of -0.1 (95% CI, -0.8 to 0.6, P=.76). Additional second-line uterotonics were required in 30.0% of the methylergonovine arm and 34.0% in the carboprost arm (adjusted odds ratio 0.72, 95% CI, 0.27-1.89, P=.505), and geometric mean quantitative blood loss was 756 mL (95% CI, 636-898) and 708 mL (95% CI, 619-810) (adjusted ratio of geometric means 1.06, 95% CI, 0.86-1.31, P=.588), respectively. No differences were detected in the occurrence of other interventions for uterine atony or postpartum hemorrhage. No difference was detected in uterine tone scores 10 minutes after administration of either methylergonovine or carboprost for refractory uterine atony, indicating that either agent is acceptable. ClinicalTrials.gov, NCT03584854.

Effect of unsupported sleepers on vertical levelling loss of heavy-haul railway track geometry under cyclic loadings

With an emphasis on the combined degradation of railway track geometry and components, an improved numerical approach is proposed for predicting the track geometrical vertical levelling loss (VLL). In contrast to previous studies, this research unprecedentedly considers the influence of unsupported sleepers (US) configuration on VLL under cyclic loadings, elasto-plastic behaviour, and different operational dynamic conditions. The nonlinear numerical models are performed adopting an explicit finite element (FE) package, and their results are validated by field data. The outcomes are iteratively regressed by an analytical logarithmic function that cumulates permanent settlements, and by a power function factor, which innovatively extends the response of US on VLL over a long term. Results shows that at 3 million cycles (or 60 MGT) the worst configuration for 20-ton axle load is at 5 US with 5-mm gap (5,51%), whereas for 30 and 40-ton axle loads is at 5 US with 2-mm gap (1.23% and 0,89%, respectively). This indicates that the axle load affects considerably the VLL as expected, however, the US condition plays an important role to accelerate it. Based on this study, the acceptable configuration of US can be specified for a minimum effect on VLL (thresholds) and, therefore, supports the development of practical maintenance guidelines to prolong the railway track service life.

Constraining the Geometry of NeRFs for Accurate DSM Generation from Multi-View Satellite Images

Neural Radiance Fields (NeRFs) are an emerging approach to 3D reconstruction that use neural networks to reconstruct scenes. However, its applications for multi-view satellite photogrammetry, which aim to reconstruct the Earth’s surface, struggle to acquire accurate digital surface models (DSMs). To address this issue, a novel framework, Geometric Constrained Neural Radiance Field (GC-NeRF) tailored for multi-view satellite photogrammetry, is proposed. GC-NeRF achieves higher DSM accuracy from multi-view satellite images. The key point of this approach is a geometric loss term, which constrains the scene geometry by making the scene surface thinner. The geometric loss term alongside z-axis scene stretching and multi-view DSM fusion strategies greatly improve the accuracy of generated DSMs. During training, bundle-adjustment-refined satellite camera models are used to cast rays through the scene. To avoid the additional input of altitude bounds described in previous works, the sparse point cloud resulting from the bundle adjustment is converted to an occupancy grid to guide the ray sampling. Experiments on WorldView-3 images indicate GC-NeRF’s superiority in accurate DSM generation from multi-view satellite images.

Forest fire detection utilizing ghost Swin transformer with attention and auxiliary geometric loss

Forest fires are a devastating natural disaster. Existing fire detection models face limitations in dataset availability, multi-scale feature extraction, and locating obscured or small flames and smoke. To address these issues, we develop a dataset containing real and synthetic forest fire images, sourced from a UAV (Unmanned Aerial Vehicle) perspective. Additionally, we propose the Ghost Convolution Swin Transformer (GCST) module to extract multi-scale flame and smoke features from different receptive fields by integrating parallel Ghost convolution and Swin Transformer. Subsequently, we design a lightweight reparameterized rotation attention module, which captures interactions across channel and spatial dimensions to suppress background noise and focus on obscured flames and smoke. Finally, we introduce a loss function, called Efficient Auxiliary Geometric Intersection over Union (EAGIoU), which employs an auxiliary bounding box to accelerate the model's convergence while integrating the geometrical principles of the predicted and real bounding boxes to accurately locate small flames and smoke. Extensive experimental results demonstrate that our method achieves 75.2 % mAP@0.5 and 42 % mAP@0.5:0.95 with a frame rate of 239 frames per second, indicating a significant improvement in accuracy and real-time performance compared to state-of-the-art techniques. The code and datasets are available at https://github.com/luckylil/forest-fire-detection.

Secrecy performance analysis of a IRS-assisted underwater optical wireless communication system

In response to the dearth of radio frequency (RF) equivalents, there has been a recent surge in interest in optical wireless communication in underwater environments. To ensure a strong line-of-sight (LOS) connection, the intelligent reflecting surface (IRS) is installed to create a virtual LOS. Then, the first part of this study looks into the security of underwater optical wireless communication (UOWC) in relation to a number of real-world phenomena, including oceanic propagation loss, oceanic turbulence, and IRS-induced geometric loss. Then, a wiretap channel with three authorized users - a reputable broadcaster named Alice (the submarine), a law-abiding user named Bob, and an eavesdropper named Eve - is examined over turbulent channels that exhibit the Log-normal distribution. Furthermore, our study derives the closed-form formulas for the secrecy performance measures, secrecy outage probability, and secrecy throughput. Finally, the numerical results show how the impact of oceanic turbulence-induced fading and distance between Bob’s and Eve’s positions on the secrecy system performance.

Non-linear influences of track dynamic irregularities on vertical levelling loss of heavy-haul railway track geometry under cyclic loadings

Abstract With an emphasis on the combined degradation of railway track geometry and components, a new numerical-analytical method is proposed for predicting the track geometrical vertical levelling loss (VLL). In contrast to previous studies, this research unprecedentedly considers the influence of initial track irregularities (ITI) on VLL under cyclic loadings, elastic-plastic behaviour, and different operational dynamic conditions. The non-linear numerical models are simulated using an explicit finite element package known as LS-Dyna, and their results are validated by full-scale experimental and field measurement data. The outcomes are iteratively regressed by an analytical logarithmic function that cumulates permanent settlements, which innovatively extends the effect of ITI on VLL in a long-term behaviour. For a typical heavy-haul railway operating under 30 tons axle load and 60 km/h train velocity, the result indicates that the set of ITI with the highest standard deviation (SD) of vertical profile (VP) degrades faster (37% on average) than that with the lowest SD. Additionally, our new findings reveal that the worst scenario is related to a train running at 60 km/h and carrying a load of 20 tons/axle in an uneven track whose SD of VP evolves from 3.23 mm at N = 0 (ITI) to 7.20 mm, whereas the best one corresponds to a train at 60 km/h and 30 ton axle load in an uneven track whose SD of VP downgrades from 0.48 to 1.50 mm, both at 3 M cycles (or 60 million gross tons). These findings indicate the importance of considering the ITI for predicting track geometrical VLL under cyclic loadings. Therefore, based on this research, an acceptable condition (thresholds) of ITI can be redefined for a minimum effect on VLL, which can support the development of practical maintenance guidelines to extend the railway track service life.

Impact of string connection and contact defects on electrical current distribution in solar cells and modules: A model validated by magnetic field imaging

AbstractModeling of solar modules and their components is essential to quantify geometrical, optical, and electrical losses and to improve the designs and technologies in terms of performance. In most loss analysis models, the current share among the busbars of the solar cell is assumed to be equal since a symmetrical distribution of the metallization is given. The impact of string terminal connection on the current distribution among the ribbons and the resulting changes in ohmic losses has not been studied yet. In this study, a MATLAB model is developed to consider the impact of the string connector terminal position on the current distribution and the ohmic losses in the ribbons and in string connector. The model allows for the analysis of the impact of contact defects scenarios in ribbons and string connectors on the current distribution. Results show that the highest current flows at the closest busbar to the string connector terminal while the current decreases at the busbars farther away from the terminal due to higher ohmic resistance of the current path. The higher the ohmic resistance of the string connector, the more inhomogeneous the current share at busbars. Simulating a 9 busbar M6 half‐cell with 1 × 0.08 mm2 string connector, positioning the string connector terminal at the leftmost or rightmost ribbon results in 0.4 W less power compared to center connection configuration, where the string connector terminal is positioned at the center ribbon. Furthermore, simulation results show that inhomogeneity of current causes about 2.1% reduction in module power compared to the case of evenly distributed cell current, considering a 120‐haf‐cell module with the same string connector. Regarding contact defect analysis, exemplary simulations show the impact of the position of detached ribbons on the power or efficiency loss. Considering left or right connection configuration, detaching the leftmost or rightmost ribbon results in higher power loss compared to other ribbons. Detaching one cell ribbon completely from the string connector results in about 0.2%abs decrease in cell efficiency, while detaching the outer ribbon along all strings of a 120‐half‐cell module results in power loss of about 0.8%. The developed model is validated by performing magnetic field imaging (MFI) measurements, in which the magnetic flux density induced by the current carried by the ribbons is measured.

Realization of 4 × 200 Gbps 4-QAM OFDM-OWC System Using Higher Order OAM Modes for HAP-to-Satellites Scenario

Recently, there has been an increase in interest in using optical wireless communication (OWC) links on high-altitude platforms (HAPs) for satellite applications. We implement an orbital angular momentum (OAM) multiplexed orthogonal frequency division multiplexing (OFDM) system using an OWC link. A space-to-air scenario is considered in which transmission pointing errors, geometric loss, turbulence, and additional link losses are taken into account to extend the transmission range, system capacity, and throughput. At 200 Gbps per channel data rate, four different OAM modes are implemented with higher order Laguerre–Gaussian (LG) modes of [0,0], [0,13], [0,40], and [0,80]. An aggregate 800 Gbps data rate can deliver a maximum OWC range of 3300–5000 km for all channels. The maximum received power over the 1000 km range is −19.34 to −32.59 dBm with an additional gain of 0–2.5 dB. It is also possible to obtain a better performance over large distances of 500–3500 km with an error vector magnitude of 2.98–17.5%. Furthermore, a high gain of −40.80 dB, a signal-to-noise ratio (SNR) of 55.21 dB, and an optical SNR of 67.25 dB can be achieved for varied transmitter pointing errors of 0.1 rad. As compared to other literature, this system exhibits a superior performance.

PatchDPCC: A Patchwise Deep Compression Framework for Dynamic Point Clouds

When compressing point clouds, point-based deep learning models operate points in a continuous space, which has a chance to minimize the geometric fidelity loss introduced by voxelization in preprocessing. But these methods could hardly scale to inputs with arbitrary points. Furthermore, the point cloud frames are individually compressed, failing the conventional wisdom of leveraging inter-frame similarity. In this work, we propose a patchwise compression framework called patchDPCC, which consists of a patch group generation module and a point-based compression model. Algorithms are developed to generate patches from different frames representing the same object, and more importantly, these patches are regulated to have the same number of points. We also incorporate a feature transfer module in the compression model, which refines the feature quality by exploiting the inter-frame similarity. Our model generates point-wise features for entropy coding, which guarantees the reconstruction speed. The evaluation on the MPEG 8i dataset shows that our method improves the compression ratio by 47.01% and 85.22% when compared to PCGCv2 and V-PCC with the same reconstruction quality, which is 9% and 16% better than that D-DPCC does. Our method also achieves the fastest decoding speed among the learning-based compression models.

FocalDreamer: Text-Driven 3D Editing via Focal-Fusion Assembly

While text-3D editing has made significant strides in leveraging score distillation sampling, emerging approaches still fall short in delivering separable, precise and consistent outcomes that are vital to content creation. In response, we introduce FocalDreamer, a framework that merges base shape with editable parts according to text prompts for fine-grained editing within desired regions. Specifically, equipped with geometry union and dual-path rendering, FocalDreamer assembles independent 3D parts into a complete object, tailored for convenient instance reuse and part-wise control. We propose geometric focal loss and style consistency regularization, which encourage focal fusion and congruent overall appearance. Furthermore, FocalDreamer generates high-fidelity geometry and PBR textures which are compatible with widely-used graphics engines. Extensive experiments have highlighted the superior editing capabilities of FocalDreamer in both quantitative and qualitative evaluations.

Analysis of an M-ary quadrature amplitude modulated underwater optical wireless communication system considering geometrical loss by laser, turbulence, and receiver diversity

Analysis of an M-ary quadrature amplitude modulated underwater optical wireless communication system considering geometrical loss by laser, turbulence, and receiver diversity

Spontaneously interacting qubits from Gauss-Bonnet

Building on previous constructions examining how a collection of small, locally interacting quantum systems might emerge via spontaneous symmetry breaking from a single-particle system of high dimension, we consider a larger family of geometric loss functionals and explicitly construct several classes of critical metrics which “know about qubits” (KAQ). The loss functional consists of the Ricci scalar with the addition of the Gauss-Bonnet term, which introduces an order parameter that allows for spontaneous symmetry breaking. The appeal of this method is two-fold: (i) the Ricci scalar has already been shown to have KAQ critical metrics and (ii) exact equations of motions are known for loss functionals with generic curvature terms up to two derivatives. We show that KAQ critical metrics, which are solutions to the equations of motion in the space of left-invariant metrics with fixed determinant, exist for loss functionals that include the Gauss-Bonnet term. We find that exploiting the subalgebra structure leads us to natural classes of KAQ metrics which contain the familiar distributions (GUE, GOE, GSE) for random Hamiltonians. We introduce tools for this analysis that will allow for straightfoward, although numerically intensive, extension to other loss functionals and higher-dimension systems.

Geometric Path Loss Distributions in Wireless Random Networks: Theory and Applications

Geometric Path Loss Distributions in Wireless Random Networks: Theory and Applications

MissModal: Increasing Robustness to Missing Modality in Multimodal Sentiment Analysis

Abstract When applying multimodal machine learning in downstream inference, both joint and coordinated multimodal representations rely on the complete presence of modalities as in training. However, modal-incomplete data, where certain modalities are missing, greatly reduces performance in Multimodal Sentiment Analysis (MSA) due to varying input forms and semantic information deficiencies. This limits the applicability of the predominant MSA methods in the real world, where the completeness of multimodal data is uncertain and variable. The generation-based methods attempt to generate the missing modality, yet they require complex hierarchical architecture with huge computational costs and struggle with the representation gaps across different modalities. Diversely, we propose a novel representation learning approach named MissModal, devoting to increasing robustness to missing modality in a classification approach. Specifically, we adopt constraints with geometric contrastive loss, distribution distance loss, and sentiment semantic loss to align the representations of modal-missing and modal-complete data, without impacting the sentiment inference for the complete modalities. Furthermore, we do not demand any changes in the multimodal fusion stage, highlighting the generality of our method in other multimodal learning systems. Extensive experiments demonstrate that the proposed method achieves superior performance with minimal computational costs in various missing modalities scenarios (flexibility), including severely missing modality (efficiency) on two public MSA datasets.

A bearing fault detection and remaining useful life prediction method based on a multi-branch residual feature fusion mechanism and optimized weight allocation

Within the context of rapidly progressing industrial sectors, rolling bearings have become a fundamental component across an array of mechanical systems. Their fault detection and remaining useful life (RUL) estimations are vital for ensuring industrial production safety. Yet, the understated characteristics of early-stage, minor faults in bearing degradation often escape detection. Additionally, numerous existing networks overlook the critical information embedded in multi-scale features, consequently diminishing the accuracy of predictions and classifications. The present study proposes MM-InfoGAN (multi-branch residual feature fusion and multi-objective optimization information maximization generative adversarial network), an innovative approach for intelligent fault detection and RUL prediction to address these issues. MM-InfoGAN augments the network’s ability to extract bearing fault characteristics and RUL data, employing a multi-branch residual feature fusion network structure coupled with an attention mechanism. Moreover, it refines the weight allocation strategy for geometric loss and introduces a novel loss function. This function optimizes weight distribution during the GAN’s training phase, expediting the attainment of network equilibrium. The efficacy of the comprehensive MM-InfoGAN model and its integrated modules was substantiated through comparative and ablation experiments conducted on the XJTU-SY dataset and IMS Bearing dataset.

Monocular depth estimation using self-supervised learning with more effective geometric constraints

Self-supervised learning-based depth estimation from monocular videos is a challenging yet promising way for 3D environment perception. Existing methods that use photometric consistency as supervision are often fragile in the case of textureless environments or illumination variations. In this paper, we adopt geometric constraints to improve the reliability and accuracy of monocular depth estimation. We introduce an equation derived from the two-view imaging geometry, from which we also develop a novel geometric loss function that is shown to be more effective than the epipolar constraint in self-supervised learning. In addition, we design a depth reconstruction loss by explicitly aligning the depth scale of the consecutive estimated depth maps to keep the depth scale consistent. Finally, a new self-supervised learning framework is formulated by incorporating the proposed geometric constraints into depth, pose, and optical flow estimation models. Extensive experiments on KITTI, Make3D, and NYU Depth v2 datasets show that, by adding our geometric constraints, the depth estimation model compares favorably with state-of-the-art self-supervised learning methods, and each sub-task has a significant improvement.

3D Surface Reconstruction Based on Dynamic Graph Convolutional Occupancy Network

A 3D reconstruction method based on dynamic graph convolutional occupancy networks is proposed to address the issues of texture information loss, geometric information loss after voxelization, and lack of object completeness constraints in the process of 3D reconstruction using voxel representation in a block-wise manner. By constructing a dynamic graph structure for feature extraction, the method aims at restore 3D models with fewer holes and local details. In the feature extraction stage, local pooling is employed within each point cloud block to address the problem of nonsignificant texture feature loss. To tackle the issues of geometric constraint loss and insufficient scene semantic information caused by block-wise processing, a feature fusion method between adjacent blocks is proposed to learn richer scene semantic information and long-range dependencies between points. By learning features within and between blocks, each point retains as much geometric information as possible, mitigating the problem of geometric information loss due to voxelization. During the surface generation, interpolation is used to infer the occupancy value for each point, and the Marching Cubes algorithm is employed for three-dimensional surface reconstruction. Experimental validation on object-level (ShapeNet dataset) and scene-level (Synthetic Rooms dataset, MatterPort3D dataset for real-world scenes) datasets demonstrates the effectiveness and advancement of the proposed method.

Free Space Optical Communication: An Enabling Backhaul Technology for 6G Non-Terrestrial Networks

The deployment of non-terrestrial networks (NTNs) is envisioned to achieve global coverage for 6G and beyond. In addition to space nodes, aerial NTN nodes such as high-altitude platform stations (HAPSs) and rotary-wing unmanned aerial vehicles (UAVs) could be deployed, based on the intended coverage and operational altitude requirements. NTN nodes have the potential to support both wireless access and backhauling. While the onboard base station provides wireless access for the end users, the backhauling link connects the airborne/space-borne base station to the core network. With its high data transmission capability comparable to fiber optics and its ability to operate in the interference-free optical spectrum, free space optical (FSO) communication is ideally suited to backhauling requirements in NTNs. In this paper, we present a comprehensive tutorial on airborne FSO backhauling. We first delve into the fundamentals of FSO signal transmission and discuss aspects such as geometrical loss, atmospheric attenuation, turbulence-induced fading, and pointing errors, all of which are critical for determining received signal levels and related link budget calculations. Then, we discuss the requirements of airborne backhaul system architectures, based on use cases. While single-layer backhaul systems are sufficient for providing coverage in rural areas, multi-layer designs are typically required to establish connectivity in urban areas, where line of sight (LoS) links are harder to maintain. We review physical layer design principles for FSO-based airborne links, discussing both intensity modulation/direct detection (IM/DD) and coherent modulation/coherent demodulation (CM/CD). Another critical design criteria for airborne backhauling is self-sustainability, which is further discussed in our paper. We conclude the paper by discussing current challenges and future research directions. In this context, we discuss reconfigurable intelligent surfaces (RIS) and spatial division multiplexing (SDM), for improved performance and an extended transmission range. We emphasize the importance of advanced handover techniques and scalability issues for practical implementation. We also highlight the growing role of artificial intelligence/machine learning (AI/ML) and their potential applications in the design and optimization of future FSO-based NTNs.