检索结果-内蒙古大学图书馆

Dynamic sequential neighbor processing: A liquid neural network-inspired framework for enhanced graph neural networks

学校读者我要写书评

暂无评论

Information sciences 2025年 719卷

作者： Kuijie Zhang Shanchen Pang Yuanyuan Zhang Yun Bai Luqi Wang Jerry Chun-Wei Lin State Key Laboratory of Chemical Safety Shandong Key Laboratory of Intelligent Oil & Gas Industrial Software the Qingdao Institute of Software China University of Petroleum Qingdao 266580 China School of information and control engineering Qingdao University of Technology Shandong266520 China Faculty of Automatic Control Electronics and Computer Science Department of Distributed Systems and IT Devices Silesian University of Technology 44-100 Gliwice Poland

Integrating information from multi-order neighborhoods is a fundamental strategy in Graph Neural Networks (GNNs) for capturing higher-order structural patterns and enhancing the expressive power of node representations. However, most existing GNNs treat neighbors from different orders as unordered sets and integrate them using static or parallel strategies, thus overlooking the sequential and evolving nature of neighborhood expansion. To address this limitation, we propose a novel GNN framework, SL, which integrates Serialized Neighbor Features with Liquid Neural Networks (LNNs) to enable order-aware, dynamic adaptation of neighbor influence. By modeling neighbor features as ordered sequences and leveraging LNNs' internal feedback dynamics, SL adapts feature extraction in real time based on local context and propagation history. This design offers fine-grained control over hierarchical dependencies and allows dynamic modulation of contributions from different neighborhood layers. SL is model-agnostic and can be seamlessly integrated with both classical and state-of-the-art GNNs. Extensive experiments across ten benchmark datasets show that SL consistently improves node classification accuracy and significantly alleviates over-smoothing in deep GNNs. These results highlight that order-aware and dynamically regulated propagation represents a powerful, flexible alternative to traditional multi-order aggregation, enhancing the adaptability and expressiveness of GNNs for complex graph learning tasks.

关键词：

Output regulation for load frequency control

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Silani, Amirreza Cucuzzella, Michele Scherpen, Jacquelien M.A. Yazdanpanah, Mohammad Javad Jan C. Wilems Center for Systems and Control ENTEG Faculty of Science and Engineering University of Groningen Nijenborgh 4 Groningen9747 AG Netherlands Control & Intelligent Processing Center of Excellence School of Electrical and Computer Engineering University of Tehran Tehran Iran

Motivated by the inadequacy of the existing control strategies for power systems affected by time-varying uncontrolled power injections such as loads and the increasingly widespread renewable energy sources, this paper proposes two control schemes based on the well-known output regulation control methodology. The first one is designed based on the classical output regulation theory and addresses the so-called Load Frequency Control (LFC) problem in presence of time-varying uncontrolled power injections. Then, in order to also minimize the generation costs, we use an approximate output regulation method that solves numerically only the partial differential equation of the regulator equation and propose a controller based on this solution, minimizing an appropriate penalty function. An extensive case study shows excellent performance of the proposed control schemes in different and critical scenarios. Copyright © 2020, The Authors. All rights reserved.

关键词： Scheduling

Necessary and sufficient conditions for the dynamic output feedback stabilization of fractional-order systems with order 0 < α < 1

学校读者我要写书评

暂无评论

science China(Information sciences) 2019年第9期62卷 205-207页

作者： Ying GUO Chong LIN Bing CHEN Qingguo WANG Institute of Complexity Science Qingdao University School of Mathematics and Statistics Zaozhuang University Institute for Intelligent Systems Faculty of Engineering and the Built EnvironmentUniversity of Johannesburg

Dear editor,Fractional calculus is the generalization of integer order calculus that enables us to obtain a considerably accurate description of several real world systems because of its memory property and practical applications. The fractional-order systems(FOS) have attracted significant attention in recent years and have yielded several valuable results, especially with respect to stability and sta-

关键词： FOS Necessary and sufficient conditions for the dynamic output feedback stabilization of fractional-order systems with order 0

Domain Adaptation and Image Classification via Deep Conditional Adaptation Network

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Ge, Pengfei Ren, Chuan-Xian Dai, Dao-Qing Yan, Hong The School of Mathematics and Systems Science Guangdong Polytechnic Normal University Guangzhou China The Intelligent Data Center School of Mathematics Sun Yat-Sen University Guangzhou China The Department of Electrical Engineering City University of Hong Kong Hong Kong Hong Kong

Unsupervised domain adaptation (UDA) aims to generalize the supervised model trained on a source domain to an unlabeled target domain. Feature space alignment-based methods are widely used to reduce the domain discrepancy between the source and target domains. However, existing feature space alignment-based methods often ignore the conditional dependence between features and labels when aligning feature spaces, which may suffer from negative transfer. In this paper, we propose a novel UDA method, Deep Conditional Adaptation Network (DCAN), based on conditional distribution alignment of feature spaces. To be specific, we reduce the domain discrepancy by minimizing the Conditional Maximum Mean Discrepancy between the conditional distributions of deep features on the source and target domains, and extract the discriminant information from target domain by maximizing the mutual information between samples and the prediction labels. In addition, DCAN can be used to address a special scenario, Partial UDA, where the target domain category is a subset of the source domain category. Experiments on both UDA and Partial UDA show that DCAN achieves superior classification performance over state-of-the-art methods. Copyright © 2020, The Authors. All rights reserved.

关键词： Image classification

Probabilistic Margins for Instance Reweighting in Adversarial Training

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Wang, Qizhou Liu, Feng Han, Bo Liu, Tongliang Gong, Chen Niu, Gang Zhou, Mingyuan Sugiyama, Masashi Department of Computer Science Hong Kong Baptist University Hong Kong DeSI Lab Australian Artificial Intelligence Institute University of Technology Sydney Australia TML Lab School of Computer Science Faculty of Engineering The University of Sydney Australia PCA Lab Key Lab of Intelligent Perception and Systems for High-Dimensional Information of MoE Jiangsu Key Lab of Image and Video Understanding for Social Security School of Computer Science and Engineering Nanjing University of Science and Technology China McCombs School of Business The University of Texas at Austin United States Graduate School of Frontier Sciences The University of Tokyo Japan

Reweighting adversarial data during training has been recently shown to improve adversarial robustness, where data closer to the current decision boundaries are regarded as more critical and given larger weights. However, existing methods measuring the closeness are not very reliable: they are discrete and can take only a few values, and they are path-dependent, i.e., they may change given the same start and end points with different attack paths. In this paper, we propose three types of probabilistic margin (PM), which are continuous and path-independent, for measuring the aforementioned closeness and reweighting adversarial data. Specifically, a PM is defined as the difference between two estimated class-posterior probabilities, e.g., such a probability of the true label minus the probability of the most confusing label given some natural data. Though different PMs capture different geometric properties, all three PMs share a negative correlation with the vulnerability of data: data with larger/smaller PMs are safer/riskier and should have smaller/larger weights. Experiments demonstrated that PMs are reliable and PM-based reweighting methods outperformed state-of-the-art counterparts. Copyright © 2021, The Authors. All rights reserved.

关键词： Probability

Hyperspectral image classification using Uniform Manifold Approximation and Projection with fusion deep learning network

学校读者我要写书评

暂无评论

Applied Soft Computing 2025年 180卷

作者： Chen Lou Mohammed A.A. Al-qaness Sike Ni Dalal Al-Alimi Robertas Damaševičius Saeed Hamood Alsamhi College of Physics and Electronic Information Engineering Zhejiang Normal University Jinhua 321004 China Zhejiang Institute of Photoelectronics Jinhua 321004 China Faculty of Engineering Sana’a University Sana’a 12544 Yemen Centre of Real Time Computer Systems Kaunas University of Technology Kaunas 51368 Lithuania Insight Centre for Data Analytics University of Galway Ireland School of ICT Bahrain Polytechnic Isa Town Bahrain Department of Computer Science and Engineering College of Informatics Korea University Seoul 02841 Republic of Korea

Hyperspectral images (HSI) are crucial for remote sensing applications as they provide detailed spectral information that accurately identifies and analyzes various materials and land features. HSI classification faces challenges due to large data volumes and high dimensionality. Dimensionality reduction techniques help address these problems by simplifying model computation, reducing redundancy, and improving feature selection. However, traditional methods struggle to capture nonlinear structures and local–global relationships in hyperspectral data. We propose a new multi-feature fusion classification model called the Uniform Manifold Approximation and Projection (UMAP) and Simple Attention Module (SimAM) mechanism fusion network (UMAPSAMFN). The main workflow of the model consists of several steps. The network uses UMAP to map high-dimensional HSI data into a low-dimensional space while maintaining a local and global structure. The feature data are sent to the convolutional neural network (CNN) and graph convolutional network (GCN) modules to capture pixel-level features and contextual information. These data are separately processed through multi-head attention modules to enhance the ability to represent feature information. Finally, the processed data are jointly fed into a fusion module with an attention mechanism to boost feature information and achieve deep modeling results. The experimental results on three benchmark HSI datasets show that UMAPSAMFN consistently exhibits the highest classification accuracy, with overall classification accuracies of 93.28%, 96.13%, and 97.73% on the Indian Pines, Pavia University, and Salinas datasets, respectively.

关键词： Attention mechanism Convolutional neural network Graph convolutional network Hyperspectral image classification Manifold Approximation and Projection (UMAP) Remote sensing

On values of the higher derivatives of the Barnes zeta function at non-positive integers

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Sakane, Shinpei Aoki, Miho Department of Information Systems Design and Data Science Graduate School of Natural Science and Technology Shimane University 1060 Nishikawatsu Matsue Shimane690-8504 Japan Department of Mathematics Interdisciplinary Faculty of Science and Engineering Shimane University 1060 Nishikawatsu Matsue Shimane690-8504 Japan

Let x be a complex number which has a positive real part, and w1, . . ., wN be positive rational numbers. We show that wsζN (s, x | w1, . . ., wN) can be expressed as a finite linear combination of the Hurwitz zeta functions over Q(x), where ζN (s, x | w1, . . ., wN) is the Barnes zeta function and w is a positive rational number explicitly determined by w1, . . ., wN . Furthermore, we give generalizations of Kummer’s formula on the gamma function and Koyama-Kurokawa’s formulae on the multiple gamma functions, and an explicit formula for the values at non-positive integers for higher order derivatives of the Barnes zeta function in the case that x is a positive rational number, involving the generalized Stieltjes constants and the values at positive integers of the Riemann zeta function. Our formulae also makes it possible to calculate an approximation in the case that w1, . . ., wN and x are positive real *** Codes 11M41, 11M06, 11M35, 33B15 Copyright © 2020, The Authors. All rights reserved.

关键词： Rational functions

Ethical Aspects of Computational Modelling in science, Decision Support and Communication 15th

学校读者我要写书评

暂无评论

Ethical Aspects of Computational Modelling in Science, Decis...

15th International Workshop on Artificial Life and Evolutionary Computation, WIVACE 2021

作者： Füchslin, Rudolf Marcel Ambühl, Jacques Faggian, Alessia Fellermann, Harold M. Flumini, Dandolo Geller, Armando Hanczyc, Martin M. Klinkert, Andreas Krütli, Pius Matuttis, Hans-Georg Ott, Thomas Scheidegger, Stephan Schmid, Gary Bruno Smieszek, Timo Schneider, Johannes J. Steiner, Albert Weyland, Mathias S. Institute of Applied Mathematics and Physics Zurich University of Applied Sciences Zurich Technikumstrasse 9 Winterthur8400 Switzerland European Centre for Living Technology Ca’ Bottacin Dorsoduro 3911 Calle Crosera Venice30123 Italy World Bank Group Geneva Switzerland University of Trento Trento38123 Italy School of Computing Newcastle University Science Square 1 Newcastle Upon TyneNE4 5TG United Kingdom GmbH Fraumünsterstrasse 11 Zurich8001 Switzerland Institute of Data Analysis and Process Design Zurich University of Applied Sciences Rosenstrasse 3 Winterthur8400 Switzerland Transdisciplinarity Lab ETH Zurich Zurich8092 Switzerland Department of Mechanical Engineering and Intelligent Systems The University of Electrocommunications Tokyo182-8585 Japan Institute of Computational Life Sciences Zurich University of Applied Sciences Schloss 1 Wädenswil8820 Switzerland Swiss NeuroCreations GmbH Hambergersteig 25 Zurich8008 Switzerland Muesmattweg 70 Allschwil4123 Switzerland

ISBN: (纸本)9783031239281

The development of data science, the increase of computational power, the availability of the internet infrastructure for data exchange and the urgency for an understanding of complex systems require a responsible and ethical use of computational models in science, communication and decision-making. Starting with a discussion of the width of different purposes of computational models, we first investigate the process of model construction as an interplay of theory and experimentation. We emphasise the different aspects of the tension between model variables and experimentally measurable observables. The resolution of this tension is a prerequisite for the responsible use of models and an instrumental part of using models in the scientific processes. We then discuss the impact of models and the responsibility that results from the fact that models support and may also guide experimentation. Further, we investigate the difference between computational modelling in an interdisciplinary science project and computational models as tools in transdisciplinary decision support. We regard the communication of model structures and modelling results as essential;however, this communication cannot happen in a technical manner, but model structures and modelling results must be translated into a "narrative." We discuss the role of concepts from disciplines such as literary theory, communication science, and cultural studies and the potential gains that a broader approach can obtain. Considering concepts from the liberal arts, we conclude that there is, besides the responsibility of the model author, also a responsibility of the user/reader of the modelling results. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Decision making

Adopting Feature-Weighted and Scale Adaptive for Real-Time Visual Tracking

学校读者我要写书评

暂无评论

Adopting Feature-Weighted and Scale Adaptive for Real-Time V...

2019 International Conference on Artificial Intelligence and Advanced Manufacturing, AIAM 2019

作者： Dong, Qiujie He, Xuedong Ge, Haiyan Liu, Qin Liu, Yanting Li, Ajuan Zhang, Yuren Han, Aifu Zhou, Shengzong Chinese Academy of Sciences Fujian Institute of Research on the Structure of Matter Fuzhou China School of Data Science North University of China Taiyuan China School of Intelligent Systems Engineering Sun Yat-sen University Shenzhen China College of Electrical and Electronic Engineering Shandong University of Technology Zibo China

ISBN: (纸本)9781728146911

Considering poor accuracy and robustness of single-feature object tracking, researchers have proposed numerous multiple-feature integration for tracking, including deep features and traditional hand-crafted features. In this paper, we propose a feature-weighted method of Histogram of Oriented Gradient (HOG) and color names (CN) based on correlation filters. Both features are adopted independently to obtain detection response maps, which are subsequently normalized. Finally, Euclidean distance between normalized map and desired output, which is activated by a negative exponential function, is regarded as the weight of the corresponding response map. Although using feature-weighted correlation filters for tracking has achieved great improvement in accuracy and robustness, many conventional trackers have not solved the problem of scale variation and the current mainstream scale estimation methods have existed the deficiencies of computational redundancy and fixed scale-factor. In view of this problem, firstly adopting three scale-factor to roughly judge the direction of target scale variation, and then the optimal scale-factor is acquired cyclically in this direction. The methods we proposed based on ECO-HC are tested in OTB2013, OTB100 and TC128 datasets and the results demonstrate the effectiveness of our proposed methods. © 2019 IEEE.

关键词： Exponential functions