检索结果-内蒙古大学图书馆

Reconstruction-driven Dynamic Refinement based Unsupervised Domain Adaptation for Joint Optic Disc and Cup Segmentation

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Chen, Ziyang Pan, Yongsheng Xia, Yong The National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology School of Computer Science and Engineering Northwestern Polytechnical University Xi’an710072 China The Ningbo Institute of Northwestern Polytechnical University Ningbo315048 China The School of Biomedical and Engineering ShanghaiTech University Shanghai201210 China

Glaucoma is one of the leading causes of irreversible blindness. Segmentation of optic disc (OD) and optic cup (OC) on fundus images is a crucial step in glaucoma screening. Although many deep learning models have been constructed for this task, it remains challenging to train an OD/OC segmentation model that could be deployed successfully to different healthcare centers. The difficulties mainly comes from the domain shift issue, i.e., the fundus images collected at these centers usually vary greatly in the tone, contrast, and brightness. To address this issue, in this paper, we propose a novel unsupervised domain adaptation (UDA) method called Reconstruction-driven Dynamic Refinement Network (RDR-Net), where we employ a due-path segmentation backbone for simultaneous edge detection and region prediction and design three modules to alleviate the domain gap. The reconstruction alignment (RA) module uses a variational auto-encoder (VAE) to reconstruct the input image and thus boosts the image representation ability of the network in a self-supervised way. It also uses a style-consistency constraint to force the network to retain more domain-invariant information. The low-level feature refinement (LFR) module employs input-specific dynamic convolutions to suppress the domain-variant information in the obtained low-level features. The prediction-map alignment (PMA) module elaborates the entropy-driven adversarial learning to encourage the network to generate source-like boundaries and regions. We evaluated our RDR-Net against state-of-the-art solutions on four public fundus image datasets. Our results indicate that RDR-Net is superior to competing models in both segmentation performance and generalization ability. Copyright © 2023, The Authors. All rights reserved.

关键词： Convolution

On the Role of Non-Localities in Fundamental Diagram Estimation

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Liu, Jing Zheng, Fangfang Yu, Boxi Jabari, Saif School of Transportation and Logistics National Engineering Laboratory of Integrated Transportation Big Data Application Technology National United Engineering Laboratory of Integrated and Intelligent Transportation Southwest Jiaotong University Western Hi-tech Zone Sichuan Chengdu611756 China Division of Engineering New York University Abu Dhabi Saadiyat Island PO Box 129188 Abu Dhabi United Arab Emirates

We consider the role of non-localities in speed-density data used to fit fundamental diagrams from vehicle trajectories. We demonstrate that the use of anticipated densities results in a clear classification of speed-density data into stationary and non-stationary points, namely, acceleration and deceleration regimes and their separating boundary. The separating boundary represents a locus of stationary traffic states, i.e., the fundamental diagram. To fit fundamental diagrams, we develop an enhanced cross entropy minimization method that honors equilibrium traffic physics. We illustrate the effectiveness of our proposed approach by comparing it with the traditional approach that uses local speed-density states and least squares estimation. Our experiments show that the separating boundary in our approach is invariant to varying trajectory samples within the same spatio-temporal region, providing further evidence that the separating boundary is indeed a locus of stationary traffic states. © 2023, CC BY.

关键词： Entropy

A sparse semismooth Newton based proximal majorization-minimization algorithm for nonconvex square-root-loss regression problem

学校读者我要写书评

暂无评论

The Journal of Machine Learning Research

The Journal of Machine Learning Research 2020年第1期21卷 9253-9290页

作者： Peipei Tang Chengjing Wang Defeng Sun Kim-Chuan Toh School of Computer and Computing Science Zhejiang University City College Hangzhou China School of Mathematics and National Engineering Laboratory of Integrated Transportation Big Data Application Technology Southwest Jiaotong University Chengdu China Department of Applied Mathematics The Hong Kong Polytechnic University Hung Hom Hong Kong Department of Mathematics and Institute of Operations Research and Analytics National University of Singapore Singapore

In this paper, we consider high-dimensional nonconvex square-root-loss regression problems and introduce a proximal majorization-minimization (PMM) algorithm for solving these problems. Our key idea for making the proposed PMM to be efficient is to develop a sparse semismooth Newton method to solve the corresponding subproblems. By using the Kurdyka-Łojasiewicz property exhibited in the underlining problems, we prove that the PMM algorithm converges to a d-stationary point. We also analyze the oracle property of the initial subproblem used in our algorithm. Extensive numerical experiments are presented to demonstrate the high efficiency of the proposed PMM algorithm.

关键词： nonconvex square-root regression problems proximal majorization-minimization semismooth Newton method

NSGA II for Configuration Optimization of Cold Chain Logistic Network with Fully Shared Facilities and Equipments

学校读者我要写书评

暂无评论

Journal of Network Intelligence

Journal of Network Intelligence 2023年第4期8卷 1497-1516页

作者： Yu, Qu-Bo An, Le-Yao Yu, Qing-Feng Liu, Xiao-Wei Pu, Fang-Zhu Ni, Shao-Quan School of Transportation and Logistics Southwest Jiaotong University Chengdu610031 China New Corridor Logistics Group Co. Ltd. of Sichuan Port and Shipping Investment Chengdu610000 China National Engineering Laboratory of Integrated Transportation Big Data Application Technology Southwest Jiaotong University Chengdu610031 China

In the face of multiple issues such as unreasonable allocation of cold chain logistics resources, environmental pollution, and energy crises, better resource integration, optimization, and upgrading are needed. One of the methods to address this is fully shared facilities and equipment of cold chain logistics network. This paper proposes the construction of a cold chain logistics resource sharing platform, which includes five main entities: supply nodes, demand nodes, demand node logistics companies, 4PL and 3PLs, aiming to achieve timely response to cold chain logistics demands. A unified optimization method for facility and equipment configuration throughout the cold chain logistics process is selected. A dual-objective programming model is constructed to minimize logistics companies’ operational costs and carbon emissions. The optimization is performed on various aspects such as the types and quantities of facility and equipment configurations. To solve this dual-objective programming model, the NSGA-II intelligent algorithm is employed. The results show that compared to the sharing of single types of cold chain logistics resources such as shared delivery vehicles or shared cold storage, full sharing is an effective approach to reducing logistics companies’ operational costs, carbon emissions, and idle cold chain logistics resources. It optimizes the internal allocation of cold chain logistics resources within the cold chain logistics network. © 2023.

关键词： Carbon

CrossDiff: Exploring Self-Supervised Representation of Pansharpening via Cross-Predictive Diffusion Model

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Xing, Yinghui Qu, Litao Zhang, Shizhou Zhang, Kai Zhang, Yanning Ran, Lingyan Zhang, Xiuwei Yin, Hanlin The Shaanxi Provincial Key Laboratory of Speech and Image Information Processing The National Engineering Laboratory for Integrated Aerospace-GroundOcean Big Data Application Technology School of Computer Science Northwestern Polytechnical University Xi’an710072 China The Research Development Institute Northwestern Polytechnical University in Shenzhen Shenzhen518057 China

Fusion of a panchromatic (PAN) image and corresponding multispectral (MS) image is also known as pansharpening, which aims to combine abundant spatial details of PAN and spectral information of MS. Due to the absence of high-resolution MS images, available deep-learning-based methods usually follow the paradigm of training at reduced resolution and testing at both reduced and full resolution. When taking original MS and PAN images as inputs, they always obtain sub-optimal results due to the scale variation. In this paper, we propose to explore the self-supervised representation of pansharpening by designing a cross-predictive diffusion model, named CrossDiff. It has two-stage training. In the first stage, we introduce a cross-predictive pretext task to pre-train the UNet structure based on conditional DDPM, while in the second stage, the encoders of the UNets are frozen to directly extract spatial and spectral features from PAN and MS, and only the fusion head is trained to adapt for pansharpening task. Extensive experiments show the effectiveness and superiority of the proposed model compared with state-of-the-art supervised and unsupervised methods. Besides, the cross-sensor experiments also verify the generalization ability of proposed self-supervised representation learners for other satellite’s datasets. We will release our code for reproducibility. © 2024, CC BY.

关键词： Deep learning

Multispectral Pan-sharpening via Dual-Channel Convolutional Network with Convolutional LSTM Based Hierarchical Spatial-Spectral Feature Fusion

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Wang, Dong Bai, Yunpeng Li, Ying School of Computer Science National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology Shaanxi Provincial Key Laboratory of Speech & Image Information Processing Northwestern Polytechnical University Xian China School of Computing and Information Systems University of Melbourne VIC3010 Australia

Multispectral pan-sharpening aims at producing a high resolution (HR) multispectral (MS) image in both spatial and spectral domains by fusing a panchromatic (PAN) image and a corresponding MS image. In this paper, we propose a novel dual-channel network (DCNet) framework for MS pan-sharpening. In our DCNet, the dual-channel backbone involves a spatial channel to capture spatial information with a 2D CNN, and a spectral channel to extract spectral information with a 3D CNN. This heterogeneous 2D/3D CNN architecture can minimize causing spectral information distortion, which typically happens in conventional 2D CNN models. In order to fully integrate the spatial and spectral features captured from different levels, we introduce a multi-level fusion strategy. Specifically, a spatial-spectral CLSTM (S2-CLSTM) module is proposed for fusing the hierarchical spatial and spectral features, which can effectively capture correlations among multi-level features. The S2-CLSTM module attaches two fusion ways: the intra-level fusion via bi-directional lateral connections and inter-level fusion via the cell state in the S2-CLSTM. Finally, the ideal HR-MS image is recovered by a reconstruction module. Extensive experiments have been conducted at both simulated lower scale and the original scale of real-world datasets. Compared with the state-of-the-art methods, the proposed DCNet achieves superior or competitive performance. Copyright © 2020, The Authors. All rights reserved.

关键词： Convolution

A fine-grained perspective on the robustness of global cargo ship transportation networks

学校读者我要写书评

暂无评论

Journal of Geographical Sciences 2018年第7期28卷 881-899页

作者：彭澎程诗奋陈金海廖梦迪吴琳刘希亮陆锋 State Key Laboratory of Resources and Environmental Information System Institute of Geographic Sciences and Natural Resources Research CAS University of Chinese Academy of Sciences Navigation Aids Technology Research Center of Jimei University National & Local Joint Engineering Research Center for Marine Navigation Aids Services College of Geomatics Shandong University of Science and Technology Institute of Computing Technology CAS Fujian Collaborative Innovation Center for Big Data Applications in Governments Jiangsu Center for Collaborative Innovation in Geographical Information Resource Development and Application

The robustness of cargo ship transportation networks is essential to the stability of the world trade system. The current research mainly focuses on the coarse-grained, holistic cargo ship transportation network while ignoring the structural diversity of different sub-networks. In this paper, we evaluate the robustness of the global cargo ship transportation network based on the most recent Automatic Identification System（AIS） data available. First, we subdivide three typical cargo ship transportation networks（i.e., oil tanker, container ship and bulk carrier） from the original cargo ship transportation network. Then, we design statistical indices based on complex network theory and employ four attack strategies, including random attack and three intentional attacks（i.e., degree-based attack, betweenness-based attack and flux-based attack） to evaluate the robustness of the three typical cargo ship transportation networks. Finally, we compare the integrity of the remaining ports of the network when a small proportion of ports lose their function. The results show that 1） compared with the holistic cargo ship transportation network, the fine-grain-based cargo ship transportation networks can fully reflect the pattern and process of global cargo transportation; 2） different cargo ship networks behave heterogeneously in terms of their robustness, with the container network being the weakest and the bulk carrier network being the strongest; and 3） small-scale intentional attacks may have significant influence on the integrity of the container network but a minor impact on the bulk carrier and oil tanker transportation networks. These conclusions can help improve the decision support capabilities in maritime transportation planning and emergency response and facilitate the establishment of a more reliable maritime transportation ***： The robustness of cargo ship transportation networks is essential to the stability of the world trade system. The current resear

关键词： complex network fine-grained cargo ship transportation network robustness automatic identification system

Influence of Rotation on Pedestrian Flow Considering Bipedal Features: Modeling Using a Fine Discrete Floor Field Cellular Automaton

学校读者我要写书评

暂无评论

SSRN

SSRN 2022年

作者： Fu, Zhijian Xiong, Xingwen Luo, Lin Yang, Yunjia Feng, Yujing Chen, Hua School of Transportation and Logistics Southwest Jiaotong University Chengdu China National Engineering Laboratory of Integrated Transportation Big Data Application Technology Southwest Jiaotong University Chengdu China National United Engineering Laboratory of Integrated and Intelligent Transportation Southwest Jiaotong University Chengdu China Sichuan FAST Fire Safety Assessment Co. Ltd. Chengdu China

Previous studies assume pedestrians rotate with a single axis (often the geometric center of agents in simulation) for simple implementation, which ignores the asymmetry and bipedal features of pedestrian behavior. Therefore, for a higher level of simulation realism, we propose a floor field cellular automaton simulating the rotation with a dual-axis considering bipedal features. Then, we investigated the influence of rotation on the pedestrian flow in three typical scenarios where one could frequently observe the rotation behavior. The findings reveal that the rotation behaviors significantly influence the crowd phenomena of pedestrians, including the travel and flow efficiency, the shape of arching and trajectories of the crowd, and the lane formation. The pedestrians with the dual-axis rotation flow fluently and efficiently when passing through a bottleneck or a crowd: 1) in the case of single-exit room evacuation, the dual-axis rotation improves the efficiency by 5%-150%;2) in the case of unidirectional flow in the L-shaped passage, the improved efficiency by the dual-axis rotation is about 30%;3) in the counterflow, the dual-axis rotation greatly relieves the unrealistic congestion. © 2022, The Authors. All rights reserved.

关键词： Rotation

Modeling the Aberrant Behavior of Nigerian Drivers Using the Manchester Driver Behavior Questionnaire

学校读者我要写书评

暂无评论

SSRN

SSRN 2021年

作者： Labbo, Muwaffaq Safiyanu Xinguo, Jiang School of Transportation and Logistics Southwest Jiaotong University West Park High-Tech District Chengdu611756 China National Engineering Laboratory of Integrated Transportation Big Data Application Technology West Park High-Tech District Chengdu611756 China Department of Civil Engineering Kano University of Science and Technology Wudil Kano Nigeria

The Driver Behavior Questionnaire (DBQ) is one of the most widely used tool for measuring self-reported driving style and behavior. Nigeria, which is among the top countries with high rate of accident is lacking in behavioral and safety studies. Therefore, the aim of this research was to develop a measurement model for the Nigerian Drivers, and to investigate the factor structure of the DBQ in Nigerian context. 741 Nigerian Drivers of age 18 years and above, with and without driving license participated in the questionnaire. The Confirmatory Factor analysis (CFA) was used for the development of the Model, in which 20 questions out of 30 were used. Three (H1, H2, and H3) models with different factors were hypothesized, in which the second model H2 was found to be the best performing among the three models, with Violation, Lapses and Aggression as factors. The results were in consistent with the usefulness of the DBQ in characterizing risky behaviors in Nigeria, which can be used in improving road safety. © 2021, The Authors. All rights reserved.

关键词： Factor analysis