Accurate Capture Research Articles

EPID-12. NEIGHBORHOOD SOCIOECONOMIC STATUS PREDICTS SURVIVAL IN GLIOBLASTOMA: A SINGLE CENTER COHORT STUDY

Abstract BACKGROUND Glioblastoma (GBM) is the most common primary brain malignancy. Despite recent progress in treatments, prognosis for patients remain poor. Recent literature has suggested the association of GBM outcomes and socioeconomic status. However, there is limited evidence regarding the association of neighborhood-level socioeconomic status on GBM outcomes. OBJECTIVE To investigate the impact of neighborhood-level socioeconomic status on GBM outcomes. METHODS We retrospectively reviewed all adult GBM patients at a single institution from 2008 to 2023. Patient addresses underwent geospatial analysis and Area Deprivation Index (ADI) was extracted. Univariable and multivariable survival analyses were performed. A Cox proportional hazards model was used to assess the effect of ADI and other socioeconomic variables while controlling for a priori selected clinical variables with known relevance to survival. RESULTS In total, 1492 patients met inclusion criteria. The average age at diagnosis was 62±14.2 years with a median overall survival of 12.1 [Interquartile Range (IQR) 5-22] months. The median ADI was 66 (IQR 46-84). Patients with high socioeconomic deprivation (ADI>75) were less likely to receive chemotherapy (92% vs 96%, p=.03), and had worse overall survival compared to patients with low socioeconomic deprivation (10 vs 13 months, p=.001). In the multivariable model, patients with high ADI had worse overall survival (hazard ratio [HR], 1.42; 95% confidence interval [CI], 1.16-1.73, p<.001). To account for changes in WHO guidelines, we implemented the model on patients diagnosed between 2017-2023 and findings were consistent (HR 1.34, 95%CI 1.07-1.68, p=.011). CONCLUSIONS We present the largest study reporting the effect of neighborhood-level socioeconomic status on overall survival of GBM patients. Neighborhood-level socioeconomic status provides the most accurate capture of the patient-level socioeconomic determinants of health. Patients with higher ADI have worse overall survival after controlling for other socioeconomic and clinical variables. These results suggest significant association between neighborhood-level socioeconomic status and overall survival in GBM patients.

Exploring the Influence of Masculine Norms on Suicidal Ideation and Help-Seeking Behavior.

Background: Men account for 75% of all suicide deaths in Australia. Societally dominant masculine norms have been theorized to be linked with suicidality and reduced help-seeking. However, evidence is needed to establish this relationship further. Aims: To further understand the relationships between 11 masculine norms, suicidal ideation, and mental health help-seeking behavior longitudinally in Australian males. Method: We analyzed data from a cohort of 8,214 males (aged 18-55 years), using logistic regression to examine if conformity to any of the 11 masculine norms measured by the Conformity to Masculine Norms Inventory (CMNI-22) at Wave 1 was associated with suicidal ideation and help-seeking at Wave 2. Results: Analyses revealed that being in the high conformity group for the norm of emotional control at Wave 1 was associated with higher odds of suicidal ideation longitudinally. Being in the low conformity group for the global construct of masculine norms and the specific norms of emotional control and power over women was associated with higher rates of mental health help-seeking behavior longitudinally. Limitations: The CMNI-22 scale's limited construct validity and the use of a single-item measure for suicidal ideation may have restricted the accurate capture of masculine norms and suicidal behaviors in Australian men. Conclusion: These results provide support for the contention that suicidality is a profoundly gendered phenomenon by showing an association between masculine norms and suicidal ideation in men. These norms should be a point of focus of male suicide prevention initiatives.

Multi-relation spatiotemporal graph residual network model with multi-level feature attention: A novel approach for landslide displacement prediction

Accurate prediction of landslide displacement is crucial for effective early warning of landslide disasters. While most existing prediction methods focus on time-series forecasting for individual monitoring points, there is limited research on the spatiotemporal characteristics of landslide deformation. This paper proposes a novel Multi-Relation Spatiotemporal Graph Residual Network with Multi-Level Feature Attention (MFA-MRSTGRN) that effectively improves the prediction performance of landslide displacement through spatiotemporal fusion. This model integrates internal seepage factors as data feature enhancements with external triggering factors, allowing for accurate capture of the complex spatiotemporal characteristics of landslide displacement and the construction of a multi-source heterogeneous dataset. The MFA-MRSTGRN model incorporates dynamic graph theory and four key modules: multi-level feature attention, temporal-residual decomposition, spatial multi-relational graph convolution, and spatiotemporal fusion prediction. This comprehensive approach enables the efficient analyses of multi-source heterogeneous datasets, facilitating adaptive exploration of the evolving multi-relational, multi-dimensional spatiotemporal complexities in landslides. When applying this model to predict the displacement of the Liangshuijing landslide, we demonstrate that the MFA-MRSTGRN model surpasses traditional models, such as random forest (RF), long short-term memory (LSTM), and spatial temporal graph convolutional networks (ST-GCN) models in terms of various evaluation metrics including mean absolute error (MAE = 1.27 mm), root mean square error (RMSE = 1.49 mm), mean absolute percentage error (MAPE = 0.026), and R-squared (R2 = 0.88). Furthermore, feature ablation experiments indicate that incorporating internal seepage factors improves the predictive performance of landslide displacement models. This research provides an advanced and reliable method for landslide displacement prediction.

Biobased Compostable Plastics End-of-Life: Environmental Assessment Including Carbon Footprint and Microplastic Impacts.

In this paper, we examine how traditional life-cycle assessment (LCA) for bio-based and compostable plastics overlooks issues surrounding carbon sequestration and microplastic persistence. To outline biased comparisons drawn from these omitted environmental impacts, we provide, as an example, a comparative LCA for compostable biobased vs. non-compostable fossil-based materials. In doing so we (1) demonstrate the proper way to capture carbon footprints to make fair comparisons and (2) identify the overlooked issues of microplastics and the need for non-persistent alternatives. By ensuring accurate biogenic carbon capture, key contributors to CO2 evolution are properly identified, allowing well-informed changes to formulations that can reduce the environmental impact of greenhouse gas emissions. In a complimentary manner, we summarize the growing research surrounding microplastic persistence and toxicity. We highlight the fundamental ability and the growing number of studies that show that industrial composting can completely mineralize certified compostable materials. This mineralization exists as a viable solution to combat microplastic persistence, currently an absent impact category in LCA. In summary, we propose a new paradigm in which the value proposition of biobased materials can be accurately captured while highlighting compostables as a solution for the increasing microplastic accumulation in the environment.

Performance Assessment of Satellite-Based Precipitation Products in the 2023 Summer Extreme Precipitation Events over North China

In the summer of 2023, North China experienced a rare extreme precipitation storm due to Typhoons Doksuri and Khanun, leading to significant secondary disasters and highlighting the urgent need for accurate rainfall forecasting. Satellite-based quantitative precipitation estimation (QPE) products like Integrated Multi-Satellite Retrievals for GPM (IMERG) and Global Satellite Mapping of Precipitation (GSMaP) from the Global Precipitation Measurement (GPM) Mission have great potential for enhancing forecasts, necessitating a quantitative evaluation before deployment. This study uses a dense rain gauge as a benchmark to assess the accuracy and capability of the latest version 7B IMERG and version 8 GSMaP satellite-based QPE products for the 2023 summer extreme precipitation in North China. These satellite-based QPE products include four satellite-only products, namely IMERG early run (IMERG_ER) and IMERG late run (IMERG_LR), GSMaP near-real-time (GSMaP_NRT), and GSMaP microwave-infrared reanalyzed (GSMaP_MVK), along with two gauge-corrected products, namely IMERG final run (IMERG_FR) and GSMaP gauge adjusted (GSMaP_Gauge). The results show that (1) GSMaP_MVK, IMERG_LR, and IMERG_FR effectively capture the space distribution of the extreme rainfall, with relatively high correlation coefficients (CCs) of approximately 0.77, 0.75, and 0.79. The IMERG_ER, GSMaP_NRT, and GSMaP_Gauge products exhibit a less accurate spatial pattern capture (CCs about 0.66, 0.73, and 0.67, respectively). Each of the six QPE products tends to underestimate rainfall (RBs < 0%). (2) The IMERG products surpass the corresponding GSMaP products in serial rainfall measurement. IMERG_LR demonstrates superior performance with the lowest root-mean-square error (RMSE) (about 0.38 mm), the highest CC (0.97), and less underestimation (RB about −6.37%). (3) The IMERG products at rainfall rates ≥ 30 mm/h, GSMaP_NRT and GSMaP_MVK products at rainfall rates ≥ 55 mm/h, and GSMaP_Gauge products at ≥ 40 mm/h showed marked limitations in event detection, with a near-zero probability of detection (POD) and a nearly 100% false alarm ratio (FAR). In this extreme precipitation event, caution is needed when using the IMERG and GSMaP products.

A Novel Panorama Depth Estimation Framework for Autonomous Driving Scenarios Based on a Vision Transformer.

An accurate panorama depth estimation result is crucial to risk perception in autonomous driving practice. In this paper, an innovative framework is presented to address the challenges of imperfect observation and projection fusion in panorama depth estimation, enabling the accurate capture of distances from surrounding images in driving scenarios. First, the Patch Filling method is proposed to alleviate the imperfect observation of panoramic depth in autonomous driving scenarios, which constructs a panoramic depth map based on the sparse distance data provided by the 3D point cloud. Then, in order to tackle the distortion challenge faced by outdoor panoramic images, a method for image context learning, ViT-Fuse, is proposed and specifically designed for equirectangular panoramic views. The experimental results show that the proposed ViT-Fuse reduces the estimation error by 9.15% on average in driving scenarios compared with the basic method and exhibits more robust and smoother results on the edge details of the depth estimation maps.

Predicting China’s Energy Consumption and CO2 Emissions by Employing a Novel Grey Model

The accurate prediction of China’s energy consumption and CO2 emissions is important for the formulation of energy and environmental policies and achieving carbon neutrality. This paper proposes a new weighted error evaluation criterion that emphasizes the importance of new data, thereby enabling more accurate capture of the variation characteristics of new data and reflecting the principle of new information priority. By optimizing the development coefficient, grey action, and parameters in the time response formula of the traditional GM(1,1), a novel optimized model, OGMW(1,1), is constructed. The accuracy of the new model is verified by three cases from the literature. The future trends of primary energy, oil, and coal consumption and CO2 emissions in China are predicted over the next five years. The conclusions are as follows: First, the new weighted error evaluation criteria are effective and reasonable and can indicate whether a grey model can reliably use the most recent information for modeling. Second, based on the new error evaluation criteria, the development coefficient, ash action, and parameter C in the time response function can be optimized. The results show that the optimization method is reasonable. Third, compared with the traditional models GM, GMO, and ARIMA, the OGMW(1,1) provides better simulation and prediction accuracy, and new information can be prioritized more effectively. Fourth, the forecasting results indicate that China will increase its consumption of primary energy, oil, and coal, as well as its CO2 emissions. Notably, the growth rates of primary energy and oil consumption are high at approximately 22.7% and 25%, coal consumption will increase slightly, and CO2 emissions will increase by approximately 11%.

RU-Net, Multi-Scale Feature Fusion and Transfer Learning: Unlocking the Potential of Cuffless Blood Pressure Monitoring with PPG and ECG.

This study introduces an innovative deep-learning model for cuffless blood pressure estimation using PPG and ECG signals, demonstrating state-of-the-art performance on the largest clean dataset, PulseDB. The rU-Net architecture, a fusion of U-Net and ResNet, enhances both generalization and feature extraction accuracy. Accurate multi-scale feature capture is facilitated by short-time Fourier transform (STFT) time-frequency distributions and multi-head attention mechanisms, allowing data-driven feature selection. The inclusion of demographic parameters as supervisory information further elevates performance. On the calibration-based dataset, our model excels, achieving outstanding accuracy (SBP MAE ± std: 4.49 ± 4.86 mmHg, DBP MAE ± std: 2.69 ± 3.10 mmHg), surpassing AAMI standards and earning a BHS Grade A rating. Addressing the challenge of calibration-free data, we propose a fine-tuning-based transfer learning approach. Remarkably, with only 10% data transfer, our model attains exceptional accuracy (SBP MAE ± std: 4.14 ± 5.01 mmHg, DBP MAE ± std: 2.48 ± 2.93 mmHg). This study sets the stage for the development of highly accurate and reliable wearable cuffless blood pressure monitoring devices.

Airport Express Passenger Flow Prediction Based on “Time-Feature” Cooperative Attention

Accurate prediction of airport express passenger flow is the basis for realizing intelligent, refined and efficient management and control of the airport rail transit system, and is of great significance to improving airport service levels and operational efficiency. Due to the numerous influencing factors that overlap with each other, and the complex mechanism of factors affecting passenger flow timing, accurate prediction of airport express passenger flow is extremely challenging. This paper proposes an airport express rail passenger flow prediction model based on the "time-feature" collaborative attention mechanism, which achieves accurate capture of the impact of multi-dimensional factors on airport express rail passenger flow in different time series. Experiments were conducted based on actual passenger flow data of the Beijing Capital International Airport Express Rail Link, and the results showed the effectiveness of the proposed method.

Aurora B/AIR-2 regulates sister centromere resolution and CENP-A/HCP-3 organization to prevent merotelic attachments.

During cell division, the accurate capture of sister kinetochores that are built on the centromeres of chromosomes by microtubules emanating from opposite spindle poles governs faithful chromosome segregation. To ensure sister chromatids separate correctly, sister centromeres undergo resolution to achieve bi-polar orientation prior to microtubule attachments. Failure of centromere resolution increases the frequency of merotelic attachments, with microtubules from opposite poles attaching to the same sister kinetochore, causing lagging chromosome, aneuploidy, and even cancer progression. The Aurora B-mediated tension-sensing machinery to correct erroneous kinetochore-microtubule attachments has been well studied. However, preventative mechanisms to avoid merotelic attachments that occur in the earlier mitotic stage are poorly understood. In this study, we found that inactivation of mitotic kinase Aurora B/AIR-2 increases merotelic attachments in Caenorhabditis elegans. On one hand, Aurora B/AIR-2-deficient cells exhibited a delay in the occurrence of centromere resolution and a disruption in targeting condensin II components to chromatin. On the other hand, loss of Aurora B/AIR-2 results in an increased localization of centromeric proteins CENP-A/HCP-3 and M18BP1/KNL-2 as well as the kinetochore protein MIS-12 on chromatin, which may generate ectopic kinetochores causing erroneous attachments. To conclude, this study elucidated that Aurora B/AIR-2 regulates sister centromere resolution and CENP-A/HCP-3 deposition to actively prevent merotely and chromosome instability in cells.

Artificial intelligence and informatics in neonatal resuscitation

Neonatal intensive care unit resuscitative care continually evolves and increasingly relies on data. Data driven precision resuscitation care can be enabled by leveraging informatics tools and artificial intelligence. Despite technological advancements, these data are often underutilized due to suboptimal data capture, aggregation, and low adoption of artificial intelligence and analytic tools. This review describes the fundamentals and explores the evidence behind informatics and artificial intelligence tools supporting neonatal intensive care unit resuscitative care, training and education. Key findings include the need for effective interface design for accurate data capture followed by storage and translation to wisdom using analytics and artificial intelligence tools. This review addresses the issues of data privacy, bias, liability and ethical frameworks when adopting these tools. While these emerging technologies hold great promise to improve resuscitation, further study of these applications in neonatal population and awareness of informatics and artificial intelligence principles among clinicians is imperative.

Automatic medical report generation combining contrastive learning and feature difference

The automatic medical report generation is a challenging task because it requires accurate capture and description of abnormal regions, especially for those discrepancies between patient and normal. In most cases, normal region descriptions dominate the entire medical report, and existing methods may fail to focus on abnormal regions due to data bias. Medical reports can be automatically generated by combining contrastive learning with feature difference in order to capture and describe abnormal regions effectively. By capturing discrepancy attributes between the input image and normal images, this method can provide more accurate diagnostic reports and better represent the visual features of abnormal regions. Specifically, we propose the feature difference approach to make the model focus more on abnormal regions, and on the other hand, we propose the combination of contrastive learning for enhancing the visual representation of feature difference based on the feature difference approach, thus improving the performance of the model. Experimental results on the IU-Xray and MIMIC-CXR datasets demonstrate the effectiveness of our approach.

Analyzing the inspiratory airflow unsteadiness in the respiratory tract using dynamic mode decomposition method

To enhance the understanding of airflow characteristics in the human respiratory tract, the inspiratory airflow field was simulated under both tidal and quasi-steady inspiratory flow rates at the mouth inlet using the large eddy simulation method. Special attention was paid on analyzing the inspiratory airflow unsteadiness using the dynamics mode decomposition (DMD) method based on the vorticity field and comparing it with the proper orthogonal decomposition (POD) method. The following novel findings were obtained. (1) Power spectral density indicates that the inspiratory airflow is highly turbulent in the pharynx–larynx region. The vorticity field in the upper airway is more affected by inspiratory patterns compared to turbulence fluctuations. (2) The DMD results indicate that the shear flow in the pharynx–larynx region is mainly caused by flow under low-frequency modes, while the disturbances of the jet flow are caused by flow under multiple frequency modes. Steady-state inspiratory pattern demonstrates the decay characteristics different from the tidal inspiratory pattern. (3) Compared to the POD method, which may contain multiple frequency components, the DMD decomposition yields modes with a single frequency, enabling a more accurate capture of the frequency and decay characteristics of the respiratory flow under each mode. In conclusion, this study demonstrates that the DMD method is more suitable for studying the respiratory airflow unsteadiness and further confirms the necessity of adopting clinically measured inspiratory data to investigate airflow unsteadiness.

FDADNet: Detection of Surface Defects in Wood-Based Panels Based on Frequency Domain Transformation and Adaptive Dynamic Downsampling

The detection of surface defects on wood-based panels plays a crucial role in product quality control. However, due to the complex background and low contrast of defects in wood-based panel images, features extracted by traditional deep learning methods based on spatial domain processing often contain noise and blurred boundaries, which severely affects detection performance. To address these issues, we have proposed a wood-based panel surface defect detection method based on frequency domain transformation and adaptive dynamic downsampling (FDADNet). Specifically, we designed a Multi-axis Frequency Domain Weighted Information Representation Module (MFDW), which effectively decoupled the indistinguishable low-contrast defects from the background in the transform domain. Gaussian filtering was then employed to eliminate noise and blur between the defects and the background. Additionally, to tackle the issue of scale differences in defects that led to difficulties in accurate capture, we designed an Adaptive Dynamic Convolution (ADConv) module for downsampling. This method flexibly compressed and enhanced features, effectively improving the differentiation of the features of objects of varying scales in the transform space, and ultimately achieved effective defect detection. To compensate for the lack of data, we constructed a dataset of wood-based panel surface defects, WBP-DET. The experimental results showed that the proposed FDADNet effectively improved the detection performance of wood-based panel surface defects in complex scenarios, achieving a solid balance between efficiency and accuracy.

Acquisition of Flight Data from Mini Unmanned Aerial Systems

Swift progress in unmanned aerial vehicle (UAV) system technology mandates effective real-time monitoring and accurate flight data capture for building an optimum target-based cutting-edge UAVs. In the current epoch of modern-UAV era, small size and mission-oriented UAV prototypes are increasingly pervasive. This study highlights the design of a data acquisition target-based mini-UAV and introduces a novel telemetry software solution to safe and reliable communication between mini-UAV and ground control station (GCS). The primary aim is to demystify the proposed wireless communication software solution on mini-UAV capability in obtaining and managing comprehensive flight data. While creating the appropriate target-based mini-UAV prototype architecture, the features that we aim to realize for the UAV components are addressed with a system approach and the component selection is presented in detail. The integration point of the selected components is conveyed in general terms. Thus, a guiding resource for future UAV studies will be presented. Then, the processes of obtaining the flight data of mini-UAVs will be addressed. At this point, the design logic of the required commands and background software will be conveyed. Obtaining the flight data of mini-UAVs will provide important scientific contributions to the literature in terms of aviation research and safety analysis. Obtaining the flight data of mini-UAVs will meet the data set needs for flight safety improvements and air traffic management optimization studies. A general analysis of the flight parameters obtained as a result of obtaining the flight data will be performed, the predictability of the parameters will be evaluated, and the analysis of the flight phases will be performed by presenting the relationships of the parameters according to the flight scenarios and the sections taken from the parameter set in the study. This study can be considered as a concrete example of situational monitoring of UAVs.

Dynamic convolutional time series forecasting based on adaptive temporal bilateral filtering

Time series data typically contain complex dynamic patterns, which not only include linear trends and seasonal variations but also significant nonlinear changes and complex dependencies. Currently, feature extraction methods for time series data primarily employ mixed-mode extraction within the temporal domain, neglecting the effective extraction and analysis of nonlinear characteristics between different observation points and the interdependencies between variables. To address these issues, we designed an adaptive temporal bilateral filtering module that effectively preserves and highlights the nonlinear features and patterns in time series while filtering out noise and redundant information. We also designed a nonlinear feature adaptive extraction module that integrates a gating mechanism and deformable convolutions. This design allows the model to adaptively adjust the shape of the convolutional kernels according to different time steps and conditions. This enables accurate capture and extraction of nonlinear features between various time observations and dependencies between different variables. Additionally, we use stacked convolutional layers to extract local contextual features, addressing fluctuations in local features caused by changes in data distribution in real-world scenarios. In summary, we propose DCNet, a dynamic convolutional network based on an adaptive temporal bilateral filter, and evaluated its performance on twelve real-world datasets. The results indicate that DCNet consistently achieves state-of-the-art performance in both short-term and long-term forecasting tasks, with favorable runtime efficiency.

Investigating resonance frequency in advanced composite structures as the main part of bridge construction: a multi-physics simulation approach

This paper investigates the nonlinear vibration characteristics and resonance frequencies of a graphene oxide powders (GOP) reinforced beam structure, which is in contact with fluid, serving as a crucial component in bridge construction. Utilizing a multi-physics simulation approach, this study integrates both mathematical modeling and COMSOL simulations to analyze the behavior of the GOP-reinforced beam. The fluid in the analysis is modeled as incompressible, non-viscous, and irrotational, contributing to the interaction with the beam structure. The isogeometric approach, coupled with the harmonic balance method and the arc-length technique, is employed to solve the complex nonlinear partial differential equations (PDEs). This coupled method ensures the accurate capture of the nonlinear response and resonance frequencies of the beam, which are critical for the safe design and performance of bridge structures. The findings reveal the significant influence of GOP reinforcement on the vibration and stability characteristics of the beam, highlighting its potential to enhance structural performance under dynamic loading conditions. The study also emphasizes the importance of considering fluid-structure interactions in the design process, as the fluid’s presence substantially affects the vibration behavior. This research contributes to the understanding of the dynamic behavior of advanced composite structures in fluid environments, offering valuable insights for the design and analysis of bridge components. The results demonstrate the efficacy of the proposed simulation approach in addressing the complexities of nonlinear vibrations in fluid-structure interaction systems, paving the way for future developments in civil engineering applications.

Review on synergizing the Metaverse and AI-driven synthetic data: enhancing virtual realms and activity recognition in computer vision

The Metaverse’s emergence is redefining digital interaction, enabling seamless engagement in immersive virtual realms. This trend’s integration with AI and virtual reality (VR) is gaining momentum, albeit with challenges in acquiring extensive human action datasets. Real-world activities involve complex intricate behaviors, making accurate capture and annotation difficult. VR compounds this difficulty by requiring meticulous simulation of natural movements and interactions. As the Metaverse bridges the physical and digital realms, the demand for diverse human action data escalates, requiring innovative solutions to enrich AI and VR capabilities. This need is underscored by state-of-the-art models that excel but are hampered by limited real-world data. The overshadowing of synthetic data benefits further complicates the issue. This paper systematically examines both real-world and synthetic datasets for activity detection and recognition in computer vision. Introducing Metaverse-enabled advancements, we unveil SynDa’s novel streamlined pipeline using photorealistic rendering and AI pose estimation. By fusing real-life video datasets, large-scale synthetic datasets are generated to augment training and mitigate real data scarcity and costs. Our preliminary experiments reveal promising results in terms of mean average precision (mAP), where combining real data and synthetic video data generated using this pipeline to train models presents an improvement in mAP (32.35%), compared to the mAP of the same model when trained on real data (29.95%). This demonstrates the transformative synergy between Metaverse and AI-driven synthetic data augmentation.

GTAM: a molecular pretraining model with geometric triangle awareness.

Molecular representation learning is pivotal for advancing deep learning applications in quantum chemistry and drug discovery. Existing methods for molecular representation learning often fall short of fully capturing the intricate interactions within chemical bonds of 2D topological graphs and the multifaceted effects of 3D geometric conformations. To overcome these challenges, we present a novel contrastive learning strategy for molecular representation learning, named Geometric Triangle Awareness Model (GTAM). This method integrates innovative molecular encoders for both 2D graphs and 3D conformations, enabling the accurate capture of geometric dependencies among edges in graph-based molecular structures. Furthermore, GTAM is bolstered by the development of two contrastive training objectives designed to facilitate the direct transfer of edge information between 2D topological graphs and 3D geometric conformations, enhancing the functionality of the molecular encoders. Through extensive evaluations on a range of 2D and 3D downstream tasks, our model has demonstrated superior performance over existing approaches. The test code and data of GTAM are available online at https://github.com/StellaHxy/GTAM.

Optimal design of broadband, low-directivity graded index acoustic lenses for underwater communication.

Manipulating underwater pressure waves is crucial for marine exploration, as electromagnetic signals are strongly absorbed in water. However, the multi-path phenomenon complicates the accurate capture of acoustic waves by receivers. Although graded index lenses, based on metamaterials with smoothly varying properties, successfully focus pressure waves, they tend to have high directivity, which hinders practical application. This work introduces three 2D acoustic lenses made from a metamaterial composed of solid inclusions in water. We propose an optimization scheme where the pressure dynamics is governed by Helmholtz's equation, with control parameters affecting each lens cell's density and bulk modulus. Through an appropriate cost function, the optimization encourages a broadband, low-directivity lens. The large-scale optimization is solved using the Lagrangian approach, which provides an analytical expression for the cost gradient. This scheme avoids the need for a separate discretization step, allowing the design to transition directly from the desired smooth refractive index to a practical lattice structure. As a result, the optimized lens closely aligns with real-world behavior. The homogenized numerical model is validated against finite elements, which considers acoustic-elastic coupling at the microstructure level. When homogenization holds, this approach proves to be an effective design tool for achieving broadband, low-directivity acoustic lenses.