检索结果-内蒙古大学图书馆

Computer, Big data and Artificial Intelligence (ICCBD+AI), International Conference on

作者： Tian Tian Lexin Jiang Zongjia Zhang Lili Yang Xiangju Cheng School of Civil Engineering and Transportation South China University of Technology Guangzhou China Jiujiang Vocational University of Civil Engineering and Architecture Jiujiang China School of Public Administration & Emergency Management Jinan University Guangzhou China SUSTech Academy of Finance and Economics Southern University of Science and Technology Shenzhen China Department of Statistics and Data Science Southern University of Science and Technology Shenzhen China State Key Laboratory of Subtropical Building and Urban Science South China University of Technology Guangzhou China

ISBN: (数字)9798331533991

ISBN: (纸本)9798331534004

In this paper, taking Yangshuo County, which is frequently affected by mountain floods and Lijiang River transit floods together, as an example, we utilize the high-precision flood data generated by the coupled hydrological-hydrodynamic models in a machine learning sample library and construct the maximum inundation water depth prediction model (KNN-Depth) and the maximum inundation flow rate prediction model (KNN-Speed) based on the machine learning K-nearest-neighbor algorithm (KNN) to predict the maximum inundation water depth and maximum inundation flow rate. The prediction performance of the models has been tested and examined with 17 data sets of historical floods, in which a total of six flood risk classes have been classified to carry out flood risk prediction in the study area by considering the two indicators of maximum inundation water depth and maximum inundation flow rate. The results show that: 1) The maximum inundation water depth predicted by KNN-Depth is close to the simulation results of the maximum inundation water depth of the hydrodynamic model, and the mean values of the Coefficients of Determination (R 2 ), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE) indexes of the 17 floods are 0.988, 0.127m, and 0.060m, respectively; 2) The maximum inundation flow velocity generated by KNN-Speed model is similar to the simulation results of the maximum inundation flow velocity of the hydrodynamic model, and the mean values of R 2 , RMSE and MAE indicators of the 17 floods are 0.968, 0.051m/s and 0.020m/s, respectively; 3) The running time of KNN-Depth and KNN-Speed models is 2s and 10s, respectively for the prediction; 4) The research indicates that the flood risk in Yangshuo County within 500m of the banks of the main rivers during a large flood is level 6 in general, and the flood risk in the new and old urban areas is up to level 5.

关键词： Machine learning algorithms Simulation Urban areas Predictive models Nearest neighbor methods Hydrodynamics Prediction algorithms data models Rivers Floods

来源：评论

学校读者我要写书评

暂无评论

Towards Optimal Customized Architecture for Heterogeneous Federated Learning with Contrastive Cloud-Edge Model Decoupling

arXiv

引用

arXiv 2024年

作者： Chen, Xingyan Du, Tian Wang, Mu Gu, Tiancheng Zhao, Yu Kou, Gang Xu, Changqiao Wu, Dapeng Oliver Financial Intelligence and Financial Engineering Key Laboratory of Sichuan Province Institute of Digital Economy and Interdisciplinary Science Innovation School of Computer and Artificial Intelligence Southwestern University of Finance and Economics Chengdu611130 China The State Key Laboratory of Networking and Switching Technology Beijing University of Posts and Telecommunications Beijing100876 China The Data Engineering at the Department of Computer Science City University of Hong Kong Hong Kong

Federated learning, as a promising distributed learning paradigm, enables collaborative training of a global model across multiple network edge clients without the need for central data collecting. However, the heterogeneity of edge data distribution drags the model towards the local minima, which can be distant from the global optimum. Such heterogeneity often leads to slow convergence and substantial communication overhead. To address these issues, we propose a novel federated learning framework called FedCMD, a model decoupling tailored to the Cloud-edge supported federated learning that separates deep neural networks into a body for capturing shared representations in Cloud and a personalized head for migrating data heterogeneity. Our motivation is that, by the deep investigation of the performance of selecting different neural network layers as the personalized head, we found rigidly assigning the last layer as the personalized head in current studies is not always optimal. Instead, it is necessary to dynamically select the personalized layer that maximizes the training performance by taking the representation difference between neighbor layers into account. To find the optimal personalized layer, we utilize the low-dimensional representation of each layer to contrast feature distribution transfer and introduce a Wasserstein-based layer selection method, aimed at identifying the best-match layer for personalization. Additionally, a weighted global aggregation algorithm is proposed based on the selected personalized layer for the practical application of FedCMD. Extensive experiments on ten benchmarks demonstrate the efficiency and superior performance of our solution compared with nine state-of-the-art solutions. All code and results are available at https://***/elegy112138/FedCMD. Copyright © 2024, The Authors. All rights reserved.

关键词： Learning systems

来源：评论

学校读者我要写书评

暂无评论

Mvqs: Robust Multi-View Instance-Level Cost-Sensitive Learning Method for Imbalanced data Classification

SSRN

引用

SSRN 2023年

作者： Hou, Zhaojie Tang, Jingjing Li, Yan Fu, Saiji Tian, Yingjie School of Business Administration Faculty of Business Administration Southwestern University of Finance and Economics Chengdu611130 China Institute of Big Data Southwestern University of Finance and Economics Chengdu611130 China School of Economics and Management Beijing University of Posts and Telecommunications Beijing100876 China School of Economics and Management University of Chinese Academy of Sciences Beijing100190 China Research Center on Fictitious Economy and Data Science Chinese Academy of Sciences Beijing100190 China Key Laboratory of Big Data Mining and Knowledge Management Chinese Academy of Sciences Beijing100190 China MOE Social Science Laboratory of Digital Economic Forecasts and Policy Simulation UCAS Beijing100190 China

During multi-view imbalanced learning, data is collected from different sources and class labels are heavily skewed. Multi-view imbalanced learning has been studied extensively, with two main categories: multi-view cost-sensitive learning and multi-view ensemble learning. However, those methods are subject to certain limitations: 1) omission of either consensus or complementary information, 2) complex preprocessing of multi-view ensemble learning and manual assignment of misclassification costs of multi-view cost-sensitive learning, and 3) limited ability to handle noisy data. Therefore, we aim to design a concise and unified framework to grapple with the multi-view representations, imbalanced classes and noisy samples simultaneously. Inspired by the elegant metrics of SVM-style models and asymmetric quadratic type squared error loss function (QTSELF), we propose a robust multi-view instance-level cost-sensitive support vector machine for imbalanced data classification called MVQS. The consensus regularization term and combination weight strategy are employed to fully exploit multi-view information. The QTSELF loss can not only adaptively impose instance-level penalties to samples, but also make MVQS more robust to noisy samples. We solve MVQS with the gradient descent and the alternating direction method of multipliers algorithms. The comprehensive experimental results on various datasets validate that MVQS is more competitive and robust than other benchmark approaches. © 2023, The Authors. All rights reserved.

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

CONDEN-FI: Consistency and Diversity Learning-based Multi-View Unsupervised Feature and In-stance Co-Selection

arXiv

引用

arXiv 2024年

作者： Huang, Yanyong Cai, Yuxin Wang, Dongjie Yi, Xiuwen Li, Tianrui The Joint Laboratory of Data Science and Business Intelligence School of Statistics Southwestern University of Finance and Economics Chengdu611130 China The Department of Electrical Engineering and Computer Science University of Kansas LawrenceKS66045 United States The JD Intelligent Cities Research JD Intelligent Cities Business Unit Beijing100176 China The School of Computing and Artificial Intelligence Southwest Jiaotong University Chengdu611756 China

The objective of multi-view unsupervised feature and instance co-selection is to simultaneously identify the most representative features and samples from multi-view unlabeled data, which aids in mitigating the curse of dimensionality and reducing instance size to improve the performance of downstream tasks. However, existing methods treat feature selection and instance selection as two separate processes, failing to leverage the potential interactions between the feature and instance spaces. Additionally, previous co-selection methods for multi-view data require concatenating different views, which overlooks the consistent information among them. In this paper, we propose a CONsistency and DivErsity learNing-based multi-view unsupervised Feature and Instance co-selection (CONDEN-FI) to address the above-mentioned issues. Specifically, CONDEN-FI reconstructs multiview data from both the sample and feature spaces to learn representations that are consistent across views and specific to each view, enabling the simultaneous selection of the most important features and instances. Moreover, CONDEN-FI adaptively learns a view-consensus similarity graph to help select both dissimilar and similar samples in the reconstructed data space, leading to a more diverse selection of instances. An efficient algorithm is developed to solve the resultant optimization problem, and the comprehensive experimental results on real-world datasets demonstrate that CONDEN-FI is effective compared to state-of-the-art methods. Copyright © 2024, The Authors. All rights reserved.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Adaptive Collaborative Correlation Learning-based Semi-Supervised Multi-Label Feature Selection

arXiv

引用

arXiv 2024年

作者： Huang, Yanyong Yang, Li Wang, Dongjie Li, Ke Yi, Xiuwen Lv, Fengmao Li, Tianrui Joint Laboratory of Data Science and Business Intelligence School of Statistics Southwestern University of Finance and Economics Chengdu611130 China Department of Electrical Engineering and Computer Science University of Kansas LawrenceKS66045 United States JD Intelligent Cities Research and JD Intelligent Cities Business Unit Beijing100176 China School of Computing and Artificial Intelligence Southwest Jiaotong University Chengdu611756 China

Semi-supervised multi-label feature selection has recently been developed to solve the curse of dimensionality problem in high-dimensional multi-label data with certain samples missing labels. Although many efforts have been made, most existing methods use a predefined graph approach to capture the sample similarity or the label correlation. In this manner, the presence of noise and outliers within the original feature space can undermine the reliability of the resulting sample similarity graph. It also fails to precisely depict the label correlation due to the existence of unknown labels. Besides, these methods only consider the discriminative power of selected features, while neglecting their redundancy. In this paper, we propose an Adaptive Collaborative Correlation lEarning-based Semi-Supervised Multi-label Feature Selection (Access-MFS) method to address these issues. Specifically, a generalized regression model equipped with an extended uncorrelated constraint is introduced to select discriminative yet irrelevant features and maintain consistency between predicted and ground-truth labels in labeled data, simultaneously. Then, the instance correlation and label correlation are integrated into the proposed regression model to adaptively learn both the sample similarity graph and the label similarity graph, which mutually enhance feature selection performance. Extensive experimental results demonstrate the superiority of the proposed Access-MFS over other state-of-the-art methods. Copyright © 2024, The Authors. All rights reserved.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

A note on the asymptotic behavior of a mildly unstable integer-valued AR(1) model

引用

Series statistics 2024年第5期58卷

作者： Ling Peng Shujin Xie Xiaohui Liu Fukang Zhu a School of Statistics and Data Science Jiangxi University of Finance and Economics Nanchang People's Republic of Chinab Key Laboratory of Data Science in Finance and Economics Jiangxi University of Finance and Economics Nanchang People's Republic of China b Key Laboratory of Data Science in Finance and Economics Jiangxi University of Finance and Economics Nanchang People's Republic of China c School of Mathematics Jilin University Changchun People's Republic of China

This note examines the conditional least squares (CLS) estimators for integer-valued autoregressive (INAR) models, with a focus on the INAR(1) model. Our investigation reveals that the joint limiting distribution of the CLS estimators for the autoregressive coefficient and the mean of the model errors is degenerate in the case of mildly unstable INAR(1) models with moderate deviations from unity. This degeneracy may pose challenges for statistical inference. We provide commenting notes based on the asymptotic results to address this issue. Additionally, we conduct simulation studies and analyze a real-world dataset to validate the theoretical findings.

关键词： Integer-valued autoregressive models mildly unstable conditional least squares estimator limiting distribution 62M10 62M02

来源：评论

学校读者我要写书评

暂无评论

Expectile trace regression via low-rank and group sparsity regularization

引用

Series statistics 2023年第6期57卷

作者： Ling Peng Xiangyong Tan Peiwen Xiao Zeinab Rizk Xiaohui Liu a School of Statistics and Data Science and Key Laboratory of Data Science in Finance and Economics Jiangxi University of Finance and Economics Nanchang Jiangxi People's Republic of China a School of Statistics and Data Science and Key Laboratory of Data Science in Finance and Economics Jiangxi University of Finance and Economics Nanchang Jiangxi People's Republic of Chinab Faculty of Commerce Department of Applied Mathematical Statistics Damietta University Damietta El-Gadeeda City Damietta Governorate Egypt

Trace regression has received a lot of attention due to its ability to account for matrix-type covariates, including panel data, images, and genomic microarrays as special cases. However, most of its existing research focuses on the case of mean regression. In this paper, we consider the expectile trace regression, which can provide a more diversified picture of the regression relationship at different expectiles, via the low-rank and group sparsity regularization. The upper bound for the statistical rate of convergence of the regularized estimator is established under some mild conditions. Some simulations, as well as a real data example, are also provided to illustrate the finite sample performance of the developed expectile trace regression.

关键词： Expectile trace regression low-rank upper bound convergence rate matrix-type covariates 62J99 62H12

来源：评论

学校读者我要写书评

暂无评论

SEQUENCE ENTROPY FOR AMENABLE GROUP ACTIONS

arXiv

引用

arXiv 2022年

作者： Liu, Chunlin Yan, Kesong CAS Wu Wen-Tsun Key Laboratory of Mathematics School of Mathematical Sciences University of Science and Technology of China Anhui Hefei230026 China School of Mathematics and Quantitative Economics Guangxi University of Finance and Economics Guangxi Nanning530003 China Guangxi Key Laboratory of Big Data in Finance and Economics Guangxi University of Finance and Economics Guangxi Nanning530003 China

We study the sequence entropy for amenable group actions and investigate systematically spectrum and several mixing concepts via sequence entropy both in measure-theoretic dynamical systems and topological dynamical systems. Moreover, we use sequence entropy pairs to characterize weakly mixing and null systems in the topological sense. Copyright © 2022, The Authors. All rights reserved.

关键词： Dynamical systems

来源：评论

学校读者我要写书评

暂无评论

Quantifying the intrinsic randomness in sequential measurements

arXiv

引用

arXiv 2024年

作者： Liu, Xinjian Wang, Yukun Han, Yunguang Wu, Xia Beijing Key Laboratory of Petroleum Data Mining China University of Petroleum Beijing102249 China State Key Laboratory of Cryptology P.O. Box 5159 Beijing100878 China College of Computer Science and Technology Nanjing University of Aeronautics and Astronautics Nanjing211106 China School of Information Central University of Finance and Economics Beijing100081 China

In the standard Bell scenario, when making a local projective measurement on each system component, the amount of randomness generated is restricted. However, this limitation can be surpassed through the implementation of sequential measurements. Nonetheless, a rigorous definition of random numbers in the context of sequential measurements is yet to be established, except for the lower quantification in device-independent scenarios. In this paper, we define quantum intrinsic randomness in sequential measurements and quantify the randomness in the Collins-Gisin-Linden-Massar-Popescu (CGLMP) inequality sequential scenario. Initially, we investigate the quantum intrinsic randomness of the mixed states under sequential projective measurements and the intrinsic randomness of the sequential positive-operator-valued measure (POVM) under pure states. Naturally, we rigorously define quantum intrinsic randomness under sequential POVM for arbitrary quantum states. Furthermore, we apply our method to one-Alice and two-Bobs sequential measurement scenarios, and quantify the quantum intrinsic randomness of the maximally entangled state and maximally violated state by giving an extremal decomposition. Finally, using the sequential Navascues-Pironio-Acin (NPA) hierarchy in the device-independent scenario, we derive lower bounds on the quantum intrinsic randomness of the maximally entangled state and maximally violated state. © 2024, CC BY.

关键词： Quantum entanglement

来源：评论

学校读者我要写书评

暂无评论

Exploring linguistic features and user engagement in Chinese online mental health counseling

引用

Heliyon 2024年第19期10卷 e38042页

作者： Zhang, Liyuan Liu, Dexi Li, Jing Wan, Changxuan Liu, Xiping School of Computer and Artificial Intelligence Jiangxi University of Finance and Economics JiangXi Nanchang 330013 China School of Mathematics and Computer YuZhang Normal College Jiangxi Nanchang 330013 China Jiangxi Key Laboratory of Data and Knowledge Engineering Jiangxi University of Finance and Economics Jiangxi Nanchang 330013 China School of Electronic Management Science Fujian Jiangxia University Fujian Fuzhou 350108 China

With the popularity of online mental health platforms, more individuals are seeking help and receiving social support by openly discussing their problems. Therefore, it's crucial to gain a deeper understanding of which problem disclosures and social support on these platforms can attract more user attention and engagement. Previous research has primarily focused on social media forums. Our work concentrates on the professional mental health platform, intending to understand the linguistic features present in posts that promote user engagement and interaction. We employ text mining and deep learning techniques to analyze posts consisting of 22,250 questions from help-seekers and 78,328 answers providing social support extracted from the Chinese online mental health counseling platform. Initially, we analyze the high-frequency words and topics of the questions and answers to gain insights into the primary focal points and the range of topics covered in these posts. The results indicate that work-related issues are the most concerning and troublesome for help-seekers, and the topics that users follow are approximately 8 types, including growth, family, in-love, marriage, emotions, human-relations, behavioral-therapy and career. Subsequently, we analyze the language usage in question-and-answer posts with different engagement from three aspects: vocabulary categories, linguistic style matching, and language modeling, aiming to identify which linguistic features can attract more user attention and engagement. The results reveal that high-engagement answer posts exhibit a higher degree of linguistic style matching with the corresponding questions, and the use of vocabulary categories also influences the attention and engagement of the posts. By exploring the linguistic features and patterns displayed in posts with different levels of engagement on the professional online mental health platform, this study offers deep insights into user behavior and the factors that impact

关键词： Language modeling Linguistic style matching Online mental health counseling The topics of questions and answers Vocabulary categories

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：