检索结果-内蒙古大学图书馆

A semi-supervised reject inference framework with hierarchical heterogeneous networks for credit scoring

International Journal of Forecasting 2025年第3期41卷 920-939页

作者： Chen, Liao Jia, Ning Jiao, Zhixian Zhao, Hongke Cui, Runbang Wang, Huimin School of Management and E-Business Zhejiang Gongshang University Hangzhou 310018 China College of Management and Economics Tianjin University Tianjin 300072 China Laboratory of Computation and Analytics of Complex Management Systems (CACMS) Tianjin University Tianjin 300072 China Beijing Fantastic Technology Co. Ltd Beijing 100080 China State Key Laboratory of Hydrology Water Resource and Hydraulic Engineering Hohai University Nanjing 210098 China Management Science Institute Hohai University Nanjing 210098 China

Credit scoring is a popular tool for loan assessment, i.e., deciding whether to accept or reject a loan application. Traditional research into learning for credit scoring has only applied historically accepted samples without rejected applicants whose true repayment performance is absent, thereby causing both sample selection bias and wasting data. Some methods have been proposed for inferring rejected samples but they are still affected by several open problems, especially for medium- and long-term loan applications with a higher rejection rate. In particular, the heterogeneous relationships between accepted and rejected applications have not been well studied. Moreover, the complex repayment behaviors resulting from long repayment terms may lead to poor learning performance. Thus, we propose a reject inference framework with Semi-supervised Hierarchical Heterogeneous Network (S2HN) for credit scoring. We introduce a hierarchical heterogeneous network for revealing the complex connections between accepted and rejected applications, and use prospective heterogeneous repayment patterns as auxiliary information through clustering and a two-layer prediction architecture. Extensive experiments conducted based on real-world data sets demonstrated the effectiveness of our proposed method. © 2024 International Institute of Forecasters

关键词： Credit scoring Deep learning Graph neural network Heterogeneous network Reject inference

来源：评论

学校读者我要写书评

暂无评论

Odd-even effect of the mosaic modulation period of quasiperiodic hopping on the Anderson localization in a one-dimensional lattice model

引用

Physical Review B 2025年第6期111卷 064201-064201页

作者： Yi-Cai Zhang Rong Yuan Shuwei Song Mingpeng Hu Chaofei Liu Yongjian Wang School of Physics and Materials Science Guangzhou University Guangzhou 510006 China School of Mathematical Sciences Laboratory of Mathematics and Complex Systems MOE Beijing Normal University Beijing 100875 China Heilongjiang Provincial Key Laboratory of Quantum Control Harbin University of Science and Technology Harbin 150080 China School of Science Jiangxi University of Science and Technology Ganzhou 341000 China School of Mathematics and Statistics Nanjing University of Science and Technology Nanjing 210094 China

In this study, we investigate Anderson localization in a one-dimensional lattice with mosaic off-diagonal quasiperiodic hopping. Our findings reveal that the localization behavior of zero-energy states is highly dependent on the parity of the mosaic modulation period, denoted as κ. Specifically, when κ is an odd integer, there is no Anderson localization transition even for large quasiperiodic hopping strengths, and the zero-energy state remains in a critical state. On the other hand, for an even κ and generic quasiperiodic hopping, the zero-energy state becomes a localized edge state at either the left or right end of the system. Additionally, we observe that the geometric mean value of the energy spectrum is equal to the constant hopping for an even κ, while for an odd κ, it is equal to the geometric mean value of the hopping. This odd-even effect of the mosaic period also extends to other eigenstates near zero energy. More specifically, for an odd κ, an energy window in which the eigenstates remain critical even for strong quasiperiodic hopping exists. In contrast, for an even κ, an Anderson localization transition occurs as the hopping strength increases. Furthermore, we are able to accurately determine the Lyapunov exponent γ(E) and the mobility edges Ec. By analyzing the Lyapunov exponent, we identify critical regions in the hopping-energy parameter planes. Additionally, as the energy approaches the mobility edges, we observe a critical index of the localization length of ν=1. Finally, we demonstrate that different systems can be characterized by their Lyapunov exponent γ(E) and Avila's acceleration ω(E).

关键词： Lyapunov methods

来源：评论

学校读者我要写书评

暂无评论

A Novel Framework for Hierarchical Fault Detection under the Supervision of key Performance Indicators 10

A Novel Framework for Hierarchical Fault Detection under the...

引用

10th IEEE data Driven Control and Learning systems Conference, DDCLS 2021

作者： Liu, Ruijie Yang, Ying State Key Laboratory for Turbulence and Complex Systems College of Engineering Peking University Department of Mechanics and Engineering Science Beijing100871 China

ISBN: (纸本)9781665424233

This paper proposes a novel key performance indicators (KPIs) supervised hierarchical fault detection framework for model-free linear dynamic systems. In the ground process layer, a bank of residual generators are constructed with the aid of the subspace identification method, based on which the faulty process variables are isolated and the corresponding fault distribution matrix is derived. In the upper KPIs supervised layer, the dynamic relationship between the high level KPIs and the low level process variables is identified and the fault detection is realized by checking the geometric property of the fault distribution matrix transmitted from the ground layer. A numerical example is provided in the end to demonstrate the effectiveness of the proposed fault detection framework. © 2021 IEEE.

关键词： Fault detection

来源：评论

学校读者我要写书评

暂无评论

Intermittent control for nonlinear hybrid stochastic delayed systems with measurement error via dynamic event-triggered mechanism

引用

ISA Transactions 2025年

作者： Zhao, Hairui Chen, Dongyan Hu, Jun Department of Applied Mathematics Harbin University of Science and Technology Harbin150080 China Heilongjiang Provincial Key Laboratory of Optimization Control and Intelligent Analysis for Complex Systems Harbin University of Science and Technology Harbin150080 China School of Automation Harbin University of Science and Technology Harbin150080 China

This paper deals with the problem of intermittent control based on dynamic event-triggered mechanism (DETM) for nonlinear hybrid stochastic delayed systems (NHSDSs) with measurement error. The DETM and intermittent control strategy are employed to alleviate the network burden in feedback and forward channels, respectively. After giving the existence and uniqueness theorem for the solution to NHSDSs, the sufficient condition to guarantee p-th moment exponential stability and almost surely exponential stability of NHSDSs is established by comparing with the constructed auxiliary system. On this basis, the intermittent control based on self-triggered mechanism is developed. In addition, it is also proved that neither the DETM nor the self-triggered mechanism can produce the Zeno phenomenon. Finally, a simulation example is given to verify the validity of the theoretical results. © 2025 Elsevier Ltd

关键词： Stabilization

来源：评论

学校读者我要写书评

暂无评论

Effects of radiative loss on premixed planar flame propagation 38th

Effects of radiative loss on premixed planar flame propagati...

引用

38th International Symposium on Combustion, 2021

作者： Chen, Zheng State Key Laboratory for Turbulence and Complex Systems Department of Mechanics and Engineering Science College of Engineering Peking University Beijing 100871 China

Radiation is an important factor in fires. It can greatly affect large-scale flame propagation occurring in fire or explosion accidents, e.g., the flame propagating in a tunnel in the mining industry. In this study, theoretical analysis is conducted for the premixed planar flame propagation in an infinitely large domain with radiative loss. The emphasis is on the radiation-induced flow and its influence on flame propagation speed. Based on the assumptions of one-step chemistry, quasi-steady flame propagation and linear heat loss, expressions describing the change of laminar burning flux, flame propagation speed, and flame temperature with the flame position are derived. Analytical expression for flow speed is also obtained. The analytical results quantify the radiation-induced thermal and flow effects on small and large scale premixed flame propagation. It is found that the laminar burning flux and flame temperature are only affected by the radiation-induced thermal effect, while the influence on flame propagation speed is sequentially dominated by the radiation-induced thermal and flow effects. Furthermore, the theoretical results are validated by transient numerical simulations considering detailed chemistry and quartic radiative loss. Both theoretical and numerical results indicate that in fire or explosion accidents, the large-scale premixed flame propagation could be decelerated by the radiation-induced flow effect by a factor of the expansion ratio. © 2020 Elsevier Ltd.

关键词： Speed

来源：评论

学校读者我要写书评

暂无评论

Multi-Input Multi-Output Target-Speaker Voice Activity Detection For Unified, Flexible, and Robust Audio-Visual Speaker Diarization

arXiv

引用

arXiv 2024年

作者： Cheng, Ming Li, Ming The School of Computer Science Wuhan University Wuhan430072 China Suzhou Municipal Key Laboratory of Multimodal Intelligent Systems Data Science Research Center Duke Kunshan University Kunshan215316 China

Audio-visual learning has demonstrated promising results in many classical speech tasks (e.g., speech separation, automatic speech recognition, wake-word spotting). We believe that introducing visual modality will also benefit speaker diarization. To date, Target-Speaker Voice Activity Detection (TS-VAD) plays an important role in highly accurate speaker diarization. However, previous TS-VAD models take audio features and utilize the speaker’s acoustic footprint to distinguish his or her personal speech activities, which is easily affected by overlapped speech in multi-speaker scenarios. Although visual information naturally tolerates overlapped speech, it suffers from spatial occlusion, low resolution, etc. The potential modality-missing problem blocks TS-VAD towards an audio-visual approach. This paper proposes a novel Multi-Input Multi-Output Target-Speaker Voice Activity Detection (MIMO-TSVAD) framework for speaker diarization. The proposed method can take audio-visual input and leverage the speaker’s acoustic footprint or lip track to flexibly conduct audio-based, video-based, and audio-visual speaker diarization in a unified sequence-to-sequence framework. Experimental results show that the MIMO-TSVAD framework demonstrates state-of-the-art performance on the VoxConverse, DIHARD-III, and MISP 2022 datasets under corresponding evaluation metrics, obtaining the Diarization Error Rates (DERs) of 4.18%, 10.10%, and 8.15%, respectively. In addition, it can perform robustly in heavy lip-missing scenarios. Copyright © 2024, The Authors. All rights reserved.

关键词： Speech recognition

来源：评论

学校读者我要写书评

暂无评论

RTPEIR: A Reverse Trajectory Prediction Enhanced Intent Recognition Algorithm for Multi-Agent systems*

RTPEIR: A Reverse Trajectory Prediction Enhanced Intent Reco...

引用

International Conference on Control, Automation, Robotics and Vision (ICARCV)

作者： Jun-Yang Cai Qin Liu Xin-Yu Xu Kelin Lu Yang-Yang Chen Key Laboratory of Measurement and Control of Complex Systems of Engineering Ministry of Education School of Automation Southeast University Nanjing China School of Cyber Science and Engineering Southeast University Nanjing China

ISBN: (数字)9798331518493

ISBN: (纸本)9798331518509

This paper proposes a novel Reverse Trajectory Prediction Enhanced Intent Recognition Algorithm (RTPEIR) that utilizes historical data to predict earlier trajectories and enhance intent recognition accuracy. The RTPEIR algorithm incorporates two main enhancements. Firstly, it leverages a combination of past and present trajectory data to reconstruct previous agent states, thereby providing a deeper understanding of agents' strategic developments. Secondly, it integrates these reconstructed trajectories with ongoing strategy recognition processes, significantly refining the predictive accuracy. Sim-ulation experiments conducted on the StarCraft Multi-Agent Challenge (SMAC) platform show that RTPEIR outperforms existing forward trajectory prediction models by 6.5%, and existing LSTM and GRU models by 13.3% and 14.4%, respectively, in terms of intent recognition accuracy. Furthermore, when combined with deep reinforcement learning algorithms, RTPEIR demonstrates a notable improvement in win rates, highlighting its effectiveness in complex multi-agent environments.

关键词： Accuracy Intent recognition Refining Predictive models Prediction algorithms Deep reinforcement learning Trajectory Robots Long short term memory Multi-agent systems

来源：评论

学校读者我要写书评

暂无评论

Animal Shapes, Modal Analysis, and Visualization of Motion (I): Horse and Camel

引用

Journal of Geometric Analysis 2023年第10期33卷 1-60页

作者： Chen, Goong Huang, Jingtong Wei, Chunqiu Yang, Jing Scully, Matthew M. Sergeev, Alexey Chen, Ming-Chieh Krantz, Steven G. Yao, Pengfei Guo, Tiexin Wang, Junmin Department of Mathematics Texas A &M University College Station 77843 TX United States Institute for Quantum Science and Engineering Texas A &M University College Station 77843 TX United States National Center for Theoretical Sciences National Tsing Hua University Hsinchu Branch Hsinchu Taiwan Department of Mathematics and Data Science Beijing University of Civil Engineering and Architecture Beijing 100044 China School of Mathematics Beijing Institute of Technology Beijing 100081 China Department of Mechanical Engineering Cheng Shiu University Kaohsiung 83347 Taiwan Department of Mathematics Washington University at St. Louis St. Louis 63130 MO United States Key Laboratory of Complex Systems and Data Science of Ministry of Education School of Mathematics Shanxi University Shanxi Taiyuan 030006 China Institute of Systems Science Academy of Mathematics and Systems Science Academia Sinica Beijing 100190 China School of Mathematics and Statistics Central South University Hunan Changsha 410083 China

Eigenfunctions and eigenvalues of physical systems and engineering structures can reveal many of the system’s fundamental features and, therefore, become a basis for the study of inverse problems. In this series of papers, we take a reverse, direct-problem point of view;namely, given the shapes of animals, can we see the patterns of their motions or behaviors through their eigenmode analysis? This modal analysis, we believe, has never been done for living animals. Our modal analysis emphasizes dynamics, which is achieved by visualization through video animation by incorporating the time-harmonic dependence of the eigenmodes. Furthermore, we intend our modal analysis to be more realistic by encompassing the situation of the presence of a floor. Certain physical interpretations of the motion patterns from modal analysis are made. In addition, by visualization, one can see that symmetry plays an important role in motion patterns. One of our main conclusions is that shapes alone can usefully reflect or explain some animal’s behavior or motion patterns. © 2023, Mathematica Josephina, Inc.

关键词： Animal shapes CAD Camel Horse Modal analysis Motion patterns

来源：评论

学校读者我要写书评

暂无评论

Parameter Inversion of High-Dimensional Chaotic systems Using Neural Ordinary Differential Equations

Parameter Inversion of High-Dimensional Chaotic Systems Usin...

引用

data Driven Control and Learning systems (DDCLS)

作者： Hanlin Bian Wei Zhu Zhang Chen L. Jingsui Chao Pei Key Laboratory of Intelligent Analysis and Decision on Complex Systems Chongqing University of Posts and Telecommunications Chongqing China School of Computer Science and Technology Chongqing University of Posts and Telecommunications Chongqing China

ISBN: (数字)9798350361674

ISBN: (纸本)9798350361681

High-dimensional chaotic systems possess complex structural characteristics, which have potential research value in fields like encrypted communication and chaos control. The inverse problem of parameters in high-dimensional chaotic systems is of significant importance for a profound understanding of the dynamic properties of such systems. This article takes a departure from neural ordinary differential equations and embeds the structural dynamics of the dynamical systems to the parameter inversion of high-dimensional chaotic systems. Numerical experiments on different high-dimensional chaotic systems indicate that neural ordinary differential equations perform exceptionally well in solving these inversion problems, with low errors and strong robustness.

关键词： Resistance Learning systems Chaotic communication Inverse problems Noise Ordinary differential equations Robustness

来源：评论

学校读者我要写书评

暂无评论

HRSTORY: Historical News Review Based Online Story Discovery 25

HRSTORY: Historical News Review Based Online Story Discovery

引用

Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and data Mining V.1

作者： Renjie Zhou Haoran Ye Jian Wan Yong Liao Key Laboratory of Complex Systems Modeling and Simulation of the Ministry of Education Hangzhou Dianzi University Hangzhou Zhejiang China Zhejiang University of Science and Technology Hangzhou Zhejiang China University of Science and Technology of China Hefei Anhui China

ISBN: (纸本)9798400712456

Story discovery on news streams can help people quickly find story from vast amounts of news, improving the efficiency of information acquisition. Recent online story discovery methods encode text topics and then cluster articles into stories based on similarity. However, the results obtained by these methods are one-time, and clustered news cannot adaptively update in a continuous news stream. Additionally, the inadequate quality of article encoding and the presence of noise data deteriorate the performance of story discovery. To this end, we propose HRSTORY for online story discovery on news streams, which employs a historical news review method to enable news to continuously adapt to the latest environment in the stream data and make corrections and updates. Furthermore, HRSTORY captures better article embeddings through modeling multi-layer relational dependencies within the text. By using sentence-level noise masking, HRSTORY improves the relevance of news article representation to core topics and reduces the interference of noise data. Experiments on real news datasets show that HRSTORY outperforms the state-of-the-art algorithms in unsupervised online story discovery performance.

关键词： historical news

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

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

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

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

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：