Mark Franz PhD, Sara Zahedian PhD, Dhairya Parekh, Tahsin Emtenam PhD, Greg Jordan
Understanding motorist detouring behavior is critical for both traffic operations and planning applications. However, measuring real-world detouring behavior is challenging due to the need to track the movement of individual vehicles. Recent developments in high-resolution vehicle trajectory data have enabled transportation professionals to observe real-world detouring behaviors without the need to install and maintain hardware such as license plate reading cameras. This paper investigates the feasibility of vehicle probe trajectory data to capture commercial motor vehicle (CMV) detouring behavior under three unique case studies. Before doing so, a validation analysis was conducted to investigate the ability of CMV probe trajectory data to represent overall CMV volumes at well-calibrated count stations near virtual weigh stations (VWS) in Maryland. The validation analysis showed strong positive correlations (above 0.75) at all VWS stations. Upon validating the data, a methodology was applied to assess CMV detour behaviors associated with CMV enforcement activities, congestion avoidance, and incident induced temporary road closures.
{"title":"Exploring Commercial Vehicle Detouring Patterns through the Application of Probe Trajectory Data","authors":"Mark Franz PhD, Sara Zahedian PhD, Dhairya Parekh, Tahsin Emtenam PhD, Greg Jordan","doi":"arxiv-2407.17319","DOIUrl":"https://doi.org/arxiv-2407.17319","url":null,"abstract":"Understanding motorist detouring behavior is critical for both traffic\u0000operations and planning applications. However, measuring real-world detouring\u0000behavior is challenging due to the need to track the movement of individual\u0000vehicles. Recent developments in high-resolution vehicle trajectory data have\u0000enabled transportation professionals to observe real-world detouring behaviors\u0000without the need to install and maintain hardware such as license plate reading\u0000cameras. This paper investigates the feasibility of vehicle probe trajectory\u0000data to capture commercial motor vehicle (CMV) detouring behavior under three\u0000unique case studies. Before doing so, a validation analysis was conducted to\u0000investigate the ability of CMV probe trajectory data to represent overall CMV\u0000volumes at well-calibrated count stations near virtual weigh stations (VWS) in\u0000Maryland. The validation analysis showed strong positive correlations (above\u00000.75) at all VWS stations. Upon validating the data, a methodology was applied\u0000to assess CMV detour behaviors associated with CMV enforcement activities,\u0000congestion avoidance, and incident induced temporary road closures.","PeriodicalId":501309,"journal":{"name":"arXiv - CS - Computational Engineering, Finance, and Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141771818","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Luisa Vollmer, Sophie Fellenz, Fabian Jirasek, Heike Leitte, Hans Hasse
We demonstrate that thermodynamic knowledge acquired by humans can be transferred to computers so that the machine can use it to solve thermodynamic problems and produce explainable solutions with a guarantee of correctness. The actionable knowledge representation system that we have created for this purpose is called KnowTD. It is based on an ontology of thermodynamics that represents knowledge of thermodynamic theory, material properties, and thermodynamic problems. The ontology is coupled with a reasoner that sets up the problem to be solved based on user input, extracts the correct, pertinent equations from the ontology, solves the resulting mathematical problem, and returns the solution to the user, together with an explanation of how it was obtained. KnowTD is presently limited to simple thermodynamic problems, similar to those discussed in an introductory course in Engineering Thermodynamics. This covers the basic theory and working principles of thermodynamics. KnowTD is designed in a modular way and is easily extendable.
{"title":"KnowTD-An Actionable Knowledge Representation System for Thermodynamics","authors":"Luisa Vollmer, Sophie Fellenz, Fabian Jirasek, Heike Leitte, Hans Hasse","doi":"arxiv-2407.17169","DOIUrl":"https://doi.org/arxiv-2407.17169","url":null,"abstract":"We demonstrate that thermodynamic knowledge acquired by humans can be\u0000transferred to computers so that the machine can use it to solve thermodynamic\u0000problems and produce explainable solutions with a guarantee of correctness. The\u0000actionable knowledge representation system that we have created for this\u0000purpose is called KnowTD. It is based on an ontology of thermodynamics that\u0000represents knowledge of thermodynamic theory, material properties, and\u0000thermodynamic problems. The ontology is coupled with a reasoner that sets up\u0000the problem to be solved based on user input, extracts the correct, pertinent\u0000equations from the ontology, solves the resulting mathematical problem, and\u0000returns the solution to the user, together with an explanation of how it was\u0000obtained. KnowTD is presently limited to simple thermodynamic problems, similar\u0000to those discussed in an introductory course in Engineering Thermodynamics.\u0000This covers the basic theory and working principles of thermodynamics. KnowTD\u0000is designed in a modular way and is easily extendable.","PeriodicalId":501309,"journal":{"name":"arXiv - CS - Computational Engineering, Finance, and Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141771819","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
To reduce the experimental validation workload for chemical researchers and accelerate the design and optimization of high-energy-density lithium metal batteries, we aim to leverage models to automatically predict Coulombic Efficiency (CE) based on the composition of liquid electrolytes. There are mainly two representative paradigms in existing methods: machine learning and deep learning. However, the former requires intelligent input feature selection and reliable computational methods, leading to error propagation from feature estimation to model prediction, while the latter (e.g. MultiModal-MoLFormer) faces challenges of poor predictive performance and overfitting due to limited diversity in augmented data. To tackle these issues, we propose a novel method COEFF (COlumbic EFficiency prediction via Fine-tuned models), which consists of two stages: pre-training a chemical general model and fine-tuning on downstream domain data. Firstly, we adopt the publicly available MoLFormer model to obtain feature vectors for each solvent and salt in the electrolyte. Then, we perform a weighted average of embeddings for each token across all molecules, with weights determined by the respective electrolyte component ratios. Finally, we input the obtained electrolyte features into a Multi-layer Perceptron or Kolmogorov-Arnold Network to predict CE. Experimental results on a real-world dataset demonstrate that our method achieves SOTA for predicting CE compared to all baselines. Data and code used in this work will be made publicly available after the paper is published.
为了减少化学研究人员的实验验证工作量，加快高能量密度锂金属电池的设计和优化，我们旨在利用模型来自动预测基于液态电解质组成的库仑效率（CE）。现有方法中主要有两种代表性范式：机器学习和深度学习。然而，前者需要智能的输入特征选择和可靠的计算方法，从而导致从特征估计到模型预测的误差传播，而后者（如多模态-MoLFormer）由于增强数据的多样性有限，面临着预测性能差和过度拟合的挑战。为了解决这些问题，我们提出了一种新方法 COEFF（通过微调模型进行的铅效率预测），它包括两个阶段：预训练化学通用模型和在下游领域数据上进行微调。首先，我们采用公开的 MoLFormer 模型来获取电解质中每种溶剂和盐的特征向量。然后，我们对所有分子中每个标记的嵌入进行加权平均，权重由各自的电解质成分比决定。最后，我们将获得的电解质特征输入多层感知器或科尔莫哥罗夫-阿诺德网络，以预测 CE。在实际数据集上的实验结果表明，与所有基线方法相比，我们的方法在预测 CE 方面达到了 SOTA。这项工作中使用的数据和代码将在论文发表后公开。
{"title":"COEFF-KANs: A Paradigm to Address the Electrolyte Field with KANs","authors":"Xinhe Li, Zhuoying Feng, Yezeng Chen, Weichen Dai, Zixu He, Yi Zhou, Shuhong Jiao","doi":"arxiv-2407.20265","DOIUrl":"https://doi.org/arxiv-2407.20265","url":null,"abstract":"To reduce the experimental validation workload for chemical researchers and\u0000accelerate the design and optimization of high-energy-density lithium metal\u0000batteries, we aim to leverage models to automatically predict Coulombic\u0000Efficiency (CE) based on the composition of liquid electrolytes. There are\u0000mainly two representative paradigms in existing methods: machine learning and\u0000deep learning. However, the former requires intelligent input feature selection\u0000and reliable computational methods, leading to error propagation from feature\u0000estimation to model prediction, while the latter (e.g. MultiModal-MoLFormer)\u0000faces challenges of poor predictive performance and overfitting due to limited\u0000diversity in augmented data. To tackle these issues, we propose a novel method\u0000COEFF (COlumbic EFficiency prediction via Fine-tuned models), which consists of\u0000two stages: pre-training a chemical general model and fine-tuning on downstream\u0000domain data. Firstly, we adopt the publicly available MoLFormer model to obtain\u0000feature vectors for each solvent and salt in the electrolyte. Then, we perform\u0000a weighted average of embeddings for each token across all molecules, with\u0000weights determined by the respective electrolyte component ratios. Finally, we\u0000input the obtained electrolyte features into a Multi-layer Perceptron or\u0000Kolmogorov-Arnold Network to predict CE. Experimental results on a real-world\u0000dataset demonstrate that our method achieves SOTA for predicting CE compared to\u0000all baselines. Data and code used in this work will be made publicly available\u0000after the paper is published.","PeriodicalId":501309,"journal":{"name":"arXiv - CS - Computational Engineering, Finance, and Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141863337","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Financial risks trend to spread from one entity to another, ultimately leading to systemic risks. The key to preventing such risks lies in understanding the causal chains behind risk contagion. Despite this, prevailing approaches primarily emphasize identifying risks, overlooking the underlying causal analysis of risk. To address such an issue, we propose a Risk Contagion Causal Reasoning model called RC2R, which uses the logical reasoning capabilities of large language models (LLMs) to dissect the causal mechanisms of risk contagion grounded in the factual and expert knowledge embedded within financial knowledge graphs (KGs). At the data level, we utilize financial KGs to construct causal instructions, empowering LLMs to perform formal causal reasoning on risk propagation and tackle the "causal parrot" problem of LLMs. In terms of model architecture, we integrate a fusion module that aligns tokens and nodes across various granularities via multi-scale contrastive learning, followed by the amalgamation of textual and graph-structured data through soft prompt with cross multi-head attention mechanisms. To quantify risk contagion, we introduce a risk pathway inference module for calculating risk scores for each node in the graph. Finally, we visualize the risk contagion pathways and their intensities using Sankey diagrams, providing detailed causal explanations. Comprehensive experiments on financial KGs and supply chain datasets demonstrate that our model outperforms several state-of-the-art models in prediction performance and out-of-distribution (OOD) generalization capabilities. We will make our dataset and code publicly accessible to encourage further research and development in this field.
{"title":"Fusing LLMs and KGs for Formal Causal Reasoning behind Financial Risk Contagion","authors":"Guanyuan Yu, Xv Wang, Qing Li, Yu Zhao","doi":"arxiv-2407.17190","DOIUrl":"https://doi.org/arxiv-2407.17190","url":null,"abstract":"Financial risks trend to spread from one entity to another, ultimately\u0000leading to systemic risks. The key to preventing such risks lies in\u0000understanding the causal chains behind risk contagion. Despite this, prevailing\u0000approaches primarily emphasize identifying risks, overlooking the underlying\u0000causal analysis of risk. To address such an issue, we propose a Risk Contagion\u0000Causal Reasoning model called RC2R, which uses the logical reasoning\u0000capabilities of large language models (LLMs) to dissect the causal mechanisms\u0000of risk contagion grounded in the factual and expert knowledge embedded within\u0000financial knowledge graphs (KGs). At the data level, we utilize financial KGs\u0000to construct causal instructions, empowering LLMs to perform formal causal\u0000reasoning on risk propagation and tackle the \"causal parrot\" problem of LLMs.\u0000In terms of model architecture, we integrate a fusion module that aligns tokens\u0000and nodes across various granularities via multi-scale contrastive learning,\u0000followed by the amalgamation of textual and graph-structured data through soft\u0000prompt with cross multi-head attention mechanisms. To quantify risk contagion,\u0000we introduce a risk pathway inference module for calculating risk scores for\u0000each node in the graph. Finally, we visualize the risk contagion pathways and\u0000their intensities using Sankey diagrams, providing detailed causal\u0000explanations. Comprehensive experiments on financial KGs and supply chain\u0000datasets demonstrate that our model outperforms several state-of-the-art models\u0000in prediction performance and out-of-distribution (OOD) generalization\u0000capabilities. We will make our dataset and code publicly accessible to\u0000encourage further research and development in this field.","PeriodicalId":501309,"journal":{"name":"arXiv - CS - Computational Engineering, Finance, and Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141771820","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Konstantinos Vlachas, Thomas Simpson, Anthony Garland, D. Dane Quinn, Charbel Farhat, Eleni Chatzi
Reduced Order Models (ROMs) form essential tools across engineering domains by virtue of their function as surrogates for computationally intensive digital twinning simulators. Although purely data-driven methods are available for ROM construction, schemes that allow to retain a portion of the physics tend to enhance the interpretability and generalization of ROMs. However, physics-based techniques can adversely scale when dealing with nonlinear systems that feature parametric dependencies. This study introduces a generative physics-based ROM that is suited for nonlinear systems with parametric dependencies and is additionally able to quantify the confidence associated with the respective estimates. A main contribution of this work is the conditioning of these parametric ROMs to features that can be derived from monitoring measurements, feasibly in an online fashion. This is contrary to most existing ROM schemes, which remain restricted to the prescription of the physics-based, and usually a priori unknown, system parameters. Our work utilizes conditional Variational Autoencoders to continuously map the required reduction bases to a feature vector extracted from limited output measurements, while additionally allowing for a probabilistic assessment of the ROM-estimated Quantities of Interest. An auxiliary task using a neural network-based parametrization of suitable probability distributions is introduced to re-establish the link with physical model parameters. We verify the proposed scheme on a series of simulated case studies incorporating effects of geometric and material nonlinearity under parametric dependencies related to system properties and input load characteristics.
还原阶次模型（ROM）是工程领域的重要工具，因为它可以替代计算密集型数字孪生模拟器。尽管有纯数据驱动的 ROM 构建方法，但允许保留部分物理特性的方案往往会增强 ROM 的可解释性和通用性。然而，在处理具有参数依赖性特征的非线性系统时，基于物理的技术可能会对扩展产生不利影响。本研究介绍了一种基于物理的生成式 ROM，它适用于具有参数依赖性的非线性系统，此外还能量化与各自估计值相关的置信度。这项工作的一个主要贡献是将这些参数 ROM 条件化为可从监测测量中获得的特征，并以可行的在线方式进行。这与大多数现有的 ROM 方案相反，这些方案仍然局限于对基于物理的、通常是先验未知的系统参数进行规定。我们的工作利用条件变异自动编码器将所需的还原基础持续映射到从有限的输出测量中提取的特征向量上，同时还允许对 ROM 估算的相关量进行概率评估。为了重新建立与物理模型参数的联系，我们引入了一项辅助任务，即使用基于神经网络的适当概率分布参数化。我们在一系列模拟案例研究中验证了所提出的方案，其中包括与系统属性和输入负载特征相关的参数依赖性下的几何和材料非线性效应。
{"title":"A Reduced Order Model conditioned on monitoring features for estimation and uncertainty quantification in engineered systems","authors":"Konstantinos Vlachas, Thomas Simpson, Anthony Garland, D. Dane Quinn, Charbel Farhat, Eleni Chatzi","doi":"arxiv-2407.17139","DOIUrl":"https://doi.org/arxiv-2407.17139","url":null,"abstract":"Reduced Order Models (ROMs) form essential tools across engineering domains\u0000by virtue of their function as surrogates for computationally intensive digital\u0000twinning simulators. Although purely data-driven methods are available for ROM\u0000construction, schemes that allow to retain a portion of the physics tend to\u0000enhance the interpretability and generalization of ROMs. However, physics-based\u0000techniques can adversely scale when dealing with nonlinear systems that feature\u0000parametric dependencies. This study introduces a generative physics-based ROM\u0000that is suited for nonlinear systems with parametric dependencies and is\u0000additionally able to quantify the confidence associated with the respective\u0000estimates. A main contribution of this work is the conditioning of these\u0000parametric ROMs to features that can be derived from monitoring measurements,\u0000feasibly in an online fashion. This is contrary to most existing ROM schemes,\u0000which remain restricted to the prescription of the physics-based, and usually a\u0000priori unknown, system parameters. Our work utilizes conditional Variational\u0000Autoencoders to continuously map the required reduction bases to a feature\u0000vector extracted from limited output measurements, while additionally allowing\u0000for a probabilistic assessment of the ROM-estimated Quantities of Interest. An\u0000auxiliary task using a neural network-based parametrization of suitable\u0000probability distributions is introduced to re-establish the link with physical\u0000model parameters. We verify the proposed scheme on a series of simulated case\u0000studies incorporating effects of geometric and material nonlinearity under\u0000parametric dependencies related to system properties and input load\u0000characteristics.","PeriodicalId":501309,"journal":{"name":"arXiv - CS - Computational Engineering, Finance, and Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141771901","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A spatiotemporal deep learning framework is proposed that is capable of 2D full-field prediction of fracture in concrete mesostructures. This framework not only predicts fractures but also captures the entire history of the fracture process, from the crack initiation in the interfacial transition zone to the subsequent propagation of the cracks in the mortar matrix. In addition, a convolutional neural network is developed which can predict the averaged stress-strain curve of the mesostructures. The UNet modeling framework, which comprises an encoder-decoder section with skip connections, is used as the deep learning surrogate model. Training and test data are generated from high-fidelity fracture simulations of randomly generated concrete mesostructures. These mesostructures include geometric variabilities such as different aggregate particle geometrical features, spatial distribution, and the total volume fraction of aggregates. The fracture simulations are carried out in Abaqus, utilizing the cohesive phase-field fracture modeling technique as the fracture modeling approach. In this work, to reduce the number of training datasets, the spatial distribution of three sets of material properties for three-phase concrete mesostructures, along with the spatial phase-field damage index, are fed to the UNet to predict the corresponding stress and spatial damage index at the subsequent step. It is shown that after the training process using this methodology, the UNet model is capable of accurately predicting damage on the unseen test dataset by using 470 datasets. Moreover, another novel aspect of this work is the conversion of irregular finite element data into regular grids using a developed pipeline. This approach allows for the implementation of less complex UNet architecture and facilitates the integration of phase-field fracture equations into surrogate models for future developments.
{"title":"A spatiotemporal deep learning framework for prediction of crack dynamics in heterogeneous solids: efficient mapping of concrete microstructures to its fracture properties","authors":"Rasoul Najafi Koopas, Shahed Rezaei, Natalie Rauter, Richard Ostwald, Rolf Lammering","doi":"arxiv-2407.15665","DOIUrl":"https://doi.org/arxiv-2407.15665","url":null,"abstract":"A spatiotemporal deep learning framework is proposed that is capable of 2D\u0000full-field prediction of fracture in concrete mesostructures. This framework\u0000not only predicts fractures but also captures the entire history of the\u0000fracture process, from the crack initiation in the interfacial transition zone\u0000to the subsequent propagation of the cracks in the mortar matrix. In addition,\u0000a convolutional neural network is developed which can predict the averaged\u0000stress-strain curve of the mesostructures. The UNet modeling framework, which\u0000comprises an encoder-decoder section with skip connections, is used as the deep\u0000learning surrogate model. Training and test data are generated from\u0000high-fidelity fracture simulations of randomly generated concrete\u0000mesostructures. These mesostructures include geometric variabilities such as\u0000different aggregate particle geometrical features, spatial distribution, and\u0000the total volume fraction of aggregates. The fracture simulations are carried\u0000out in Abaqus, utilizing the cohesive phase-field fracture modeling technique\u0000as the fracture modeling approach. In this work, to reduce the number of\u0000training datasets, the spatial distribution of three sets of material\u0000properties for three-phase concrete mesostructures, along with the spatial\u0000phase-field damage index, are fed to the UNet to predict the corresponding\u0000stress and spatial damage index at the subsequent step. It is shown that after\u0000the training process using this methodology, the UNet model is capable of\u0000accurately predicting damage on the unseen test dataset by using 470 datasets.\u0000Moreover, another novel aspect of this work is the conversion of irregular\u0000finite element data into regular grids using a developed pipeline. This\u0000approach allows for the implementation of less complex UNet architecture and\u0000facilitates the integration of phase-field fracture equations into surrogate\u0000models for future developments.","PeriodicalId":501309,"journal":{"name":"arXiv - CS - Computational Engineering, Finance, and Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-07-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141771826","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Qihui Zhu, Shenwen Chen, Tong Guo, Yisheng Lv, Wenbo Du
Accurate flight delay prediction is crucial for the secure and effective operation of the air traffic system. Recent advances in modeling inter-airport relationships present a promising approach for investigating flight delay prediction from the multi-airport scenario. However, the previous prediction works only accounted for the simplistic relationships such as traffic flow or geographical distance, overlooking the intricate interactions among airports and thus proving inadequate. In this paper, we leverage causal inference to precisely model inter-airport relationships and propose a self-corrective spatio-temporal graph neural network (named CausalNet) for flight delay prediction. Specifically, Granger causality inference coupled with a self-correction module is designed to construct causality graphs among airports and dynamically modify them based on the current airport's delays. Additionally, the features of the causality graphs are adaptively extracted and utilized to address the heterogeneity of airports. Extensive experiments are conducted on the real data of top-74 busiest airports in China. The results show that CausalNet is superior to baselines. Ablation studies emphasize the power of the proposed self-correction causality graph and the graph feature extraction module. All of these prove the effectiveness of the proposed methodology.
{"title":"A Spatio-Temporal Approach with Self-Corrective Causal Inference for Flight Delay Prediction","authors":"Qihui Zhu, Shenwen Chen, Tong Guo, Yisheng Lv, Wenbo Du","doi":"arxiv-2407.15185","DOIUrl":"https://doi.org/arxiv-2407.15185","url":null,"abstract":"Accurate flight delay prediction is crucial for the secure and effective\u0000operation of the air traffic system. Recent advances in modeling inter-airport\u0000relationships present a promising approach for investigating flight delay\u0000prediction from the multi-airport scenario. However, the previous prediction\u0000works only accounted for the simplistic relationships such as traffic flow or\u0000geographical distance, overlooking the intricate interactions among airports\u0000and thus proving inadequate. In this paper, we leverage causal inference to\u0000precisely model inter-airport relationships and propose a self-corrective\u0000spatio-temporal graph neural network (named CausalNet) for flight delay\u0000prediction. Specifically, Granger causality inference coupled with a\u0000self-correction module is designed to construct causality graphs among airports\u0000and dynamically modify them based on the current airport's delays.\u0000Additionally, the features of the causality graphs are adaptively extracted and\u0000utilized to address the heterogeneity of airports. Extensive experiments are\u0000conducted on the real data of top-74 busiest airports in China. The results\u0000show that CausalNet is superior to baselines. Ablation studies emphasize the\u0000power of the proposed self-correction causality graph and the graph feature\u0000extraction module. All of these prove the effectiveness of the proposed\u0000methodology.","PeriodicalId":501309,"journal":{"name":"arXiv - CS - Computational Engineering, Finance, and Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-07-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141785093","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Many natural language processing (NLP) tasks in English or general domains are widely available and are often used to evaluate pre-trained language models. In contrast, there are fewer tasks available for languages other than English and for the financial domain. In particular, tasks in Japanese and the financial domain are limited. We construct two large datasets using materials published by a Japanese central government agency. The datasets provide three Japanese financial NLP tasks, which include a 3-class and 12-class classification for categorizing sentences, as well as a 5-class classification task for sentiment analysis. Our datasets are designed to be comprehensive and up-to-date, leveraging an automatic update framework that ensures the latest task datasets are publicly available anytime.
{"title":"Economy Watchers Survey provides Datasets and Tasks for Japanese Financial Domain","authors":"Masahiro Suzuki, Hiroki Sakaji","doi":"arxiv-2407.14727","DOIUrl":"https://doi.org/arxiv-2407.14727","url":null,"abstract":"Many natural language processing (NLP) tasks in English or general domains\u0000are widely available and are often used to evaluate pre-trained language\u0000models. In contrast, there are fewer tasks available for languages other than\u0000English and for the financial domain. In particular, tasks in Japanese and the\u0000financial domain are limited. We construct two large datasets using materials\u0000published by a Japanese central government agency. The datasets provide three\u0000Japanese financial NLP tasks, which include a 3-class and 12-class\u0000classification for categorizing sentences, as well as a 5-class classification\u0000task for sentiment analysis. Our datasets are designed to be comprehensive and\u0000up-to-date, leveraging an automatic update framework that ensures the latest\u0000task datasets are publicly available anytime.","PeriodicalId":501309,"journal":{"name":"arXiv - CS - Computational Engineering, Finance, and Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141771898","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Maziar Zamanpour, Suiyi He, Michael W. Levin, Zongxuan Sun
Connected and autonomous vehicles (CAVs) possess the capability of perception and information broadcasting with other CAVs and connected intersections. Additionally, they exhibit computational abilities and can be controlled strategically, offering energy benefits. One potential control strategy is real-time speed control, which adjusts the vehicle speed by taking advantage of broadcasted traffic information, such as signal timings. However, the optimal control is likely to increase the gap in front of the controlled CAV, which induces lane changing by other drivers. This study proposes a modified traffic flow model that aims to predict lane-changing occurrences and assess the impact of lane changes on future traffic states. The primary objective is to improve energy efficiency. The prediction model is based on a cell division platform and is derived considering the additional flow during lane changing. An optimal control strategy is then developed, subject to the predicted trajectory generated for the preceding vehicle. Lane change prediction estimates future speed and gap of vehicles, based on predicted traffic states. The proposed framework outperforms the non-lane change traffic model, resulting in up to 13% energy savings when lane changing is predicted 4-6 seconds in advance.
{"title":"Incorporating lane-change prediction into energy-efficient speed control of connected autonomous vehicles at intersections","authors":"Maziar Zamanpour, Suiyi He, Michael W. Levin, Zongxuan Sun","doi":"arxiv-2407.15004","DOIUrl":"https://doi.org/arxiv-2407.15004","url":null,"abstract":"Connected and autonomous vehicles (CAVs) possess the capability of perception\u0000and information broadcasting with other CAVs and connected intersections.\u0000Additionally, they exhibit computational abilities and can be controlled\u0000strategically, offering energy benefits. One potential control strategy is\u0000real-time speed control, which adjusts the vehicle speed by taking advantage of\u0000broadcasted traffic information, such as signal timings. However, the optimal\u0000control is likely to increase the gap in front of the controlled CAV, which\u0000induces lane changing by other drivers. This study proposes a modified traffic\u0000flow model that aims to predict lane-changing occurrences and assess the impact\u0000of lane changes on future traffic states. The primary objective is to improve\u0000energy efficiency. The prediction model is based on a cell division platform\u0000and is derived considering the additional flow during lane changing. An optimal\u0000control strategy is then developed, subject to the predicted trajectory\u0000generated for the preceding vehicle. Lane change prediction estimates future\u0000speed and gap of vehicles, based on predicted traffic states. The proposed\u0000framework outperforms the non-lane change traffic model, resulting in up to 13%\u0000energy savings when lane changing is predicted 4-6 seconds in advance.","PeriodicalId":501309,"journal":{"name":"arXiv - CS - Computational Engineering, Finance, and Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141771821","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Liam K. MagargalDepartment of Mechanical Engineering and Mechanics, Lehigh University, Bethlehem, PA, United States, Parisa KhodabakhshiDepartment of Mechanical Engineering and Mechanics, Lehigh University, Bethlehem, PA, United States, Steven N. RodriguezComputational Multiphysics Systems Laboratory, United States Naval Research Laboratory, Washington, DC, United States, Justin W. JaworskiKevin T. Crofton Department of Aerospace and Ocean Engineering, Virginia Tech, Blacksburg, VA, United States, John G. MichopoulosComputational Multiphysics Systems Laboratory, United States Naval Research Laboratory, Washington, DC, United States
This paper presents a graph autoencoder architecture capable of performing projection-based model-order reduction (PMOR) on advection-dominated flows modeled by unstructured meshes. The autoencoder is coupled with the time integration scheme from a traditional deep least-squares Petrov-Galerkin projection and provides the first deployment of a graph autoencoder into a PMOR framework. The presented graph autoencoder is constructed with a two-part process that consists of (1) generating a hierarchy of reduced graphs to emulate the compressive abilities of convolutional neural networks (CNNs) and (2) training a message passing operation at each step in the hierarchy of reduced graphs to emulate the filtering process of a CNN. The resulting framework provides improved flexibility over traditional CNN-based autoencoders because it is extendable to unstructured meshes. To highlight the capabilities of the proposed framework, which is named geometric deep least-squares Petrov-Galerkin (GD-LSPG), we benchmark the method on a one-dimensional Burgers' equation problem with a structured mesh and demonstrate the flexibility of GD-LSPG by deploying it to a two-dimensional Euler equations model that uses an unstructured mesh. The proposed framework provides considerable improvement in accuracy for very low-dimensional latent spaces in comparison with traditional affine projections.
{"title":"Projection-based model-order reduction for unstructured meshes with graph autoencoders","authors":"Liam K. MagargalDepartment of Mechanical Engineering and Mechanics, Lehigh University, Bethlehem, PA, United States, Parisa KhodabakhshiDepartment of Mechanical Engineering and Mechanics, Lehigh University, Bethlehem, PA, United States, Steven N. RodriguezComputational Multiphysics Systems Laboratory, United States Naval Research Laboratory, Washington, DC, United States, Justin W. JaworskiKevin T. Crofton Department of Aerospace and Ocean Engineering, Virginia Tech, Blacksburg, VA, United States, John G. MichopoulosComputational Multiphysics Systems Laboratory, United States Naval Research Laboratory, Washington, DC, United States","doi":"arxiv-2407.13669","DOIUrl":"https://doi.org/arxiv-2407.13669","url":null,"abstract":"This paper presents a graph autoencoder architecture capable of performing\u0000projection-based model-order reduction (PMOR) on advection-dominated flows\u0000modeled by unstructured meshes. The autoencoder is coupled with the time\u0000integration scheme from a traditional deep least-squares Petrov-Galerkin\u0000projection and provides the first deployment of a graph autoencoder into a PMOR\u0000framework. The presented graph autoencoder is constructed with a two-part\u0000process that consists of (1) generating a hierarchy of reduced graphs to\u0000emulate the compressive abilities of convolutional neural networks (CNNs) and\u0000(2) training a message passing operation at each step in the hierarchy of\u0000reduced graphs to emulate the filtering process of a CNN. The resulting\u0000framework provides improved flexibility over traditional CNN-based autoencoders\u0000because it is extendable to unstructured meshes. To highlight the capabilities\u0000of the proposed framework, which is named geometric deep least-squares\u0000Petrov-Galerkin (GD-LSPG), we benchmark the method on a one-dimensional\u0000Burgers' equation problem with a structured mesh and demonstrate the\u0000flexibility of GD-LSPG by deploying it to a two-dimensional Euler equations\u0000model that uses an unstructured mesh. The proposed framework provides\u0000considerable improvement in accuracy for very low-dimensional latent spaces in\u0000comparison with traditional affine projections.","PeriodicalId":501309,"journal":{"name":"arXiv - CS - Computational Engineering, Finance, and Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-07-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141745039","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：