检索结果-南通市图书馆

Transforming Education with Photogrammetry:Creating Realistic 3D Objects for Augmented Reality Applications

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Computer Modeling in Engineering & Sciences 2025年第1期142卷 185-208页

作者： Kaviyaraj Ravichandran Uma Mohan Department of Computational intelligence School of ComputingSRM Institute of Science and TechnologyKattankulathur603203India

Augmented reality(AR)is an emerging dynamic technology that effectively supports education across different *** increased use of mobile devices has an even greater *** the demand for AR applications in education continues to increase,educators actively seek innovative and immersive methods to engage students in ***,exploring these possibilities also entails identifying and overcoming existing barriers to optimal educational ***,this surge in demand has prompted the identification of specific barriers,one of which is three-dimensional(3D)*** 3D objects for augmented reality education applications can be challenging and time-consuming for the *** address this,we have developed a pipeline that creates realistic 3D objects from the two-dimensional(2D)*** for augmented and virtual reality can then utilize these created 3D *** evaluated the proposed pipeline based on the usability of the 3D object and performance ***,with 117 respondents,the co-creation team was surveyed with openended questions to evaluate the precision of the 3D object created by the proposed photogrammetry *** analyzed the survey data using descriptive-analytical methods and found that the proposed pipeline produces 3D models that are positively accurate when compared to real-world objects,with an average mean score above *** study adds new knowledge in creating 3D objects for augmented reality applications by using the photogrammetry technique;finally,it discusses potential problems and future research directions for 3D objects in the education sector.

关键词： Augmented reality education immersive learning 3D object creation photogrammetry and StructureFromMotion

Distribution-flexible subset quantization for post-quantizing super-resolution networks

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2025年第03期68卷 163-180页

作者： Yunshan ZHONG Mingbao LIN Jingjing XIE Yuxin ZHANG Fei CHAO Rongrong JI Institute of Artificial intelligence Xiamen University Key Laboratory of Multimedia Trusted Perception and Efficient computing Ministry of Education of ChinaXiamen University Tencent Youtu Lab Department of Artificial intelligence School of Informatics Xiamen University Peng Cheng Laboratory

This paper introduces distribution-flexible subset quantization(DFSQ), a post-training quantization method for super-resolution networks. Our motivation for developing DFSQ is based on the distinctive activation distributions of current super-resolution models, which exhibit significant variance across samples and channels. To address this issue, DFSQ conducts channel-wise normalization of the activations and applies distribution-flexible subset quantization(SQ), wherein the quantization points are selected from a universal set consisting of multi-word additive log-scale values. To expedite the selection of quantization points in SQ, we propose a fast quantization points selection strategy that uses K-means clustering to select the quantization points closest to the centroids. Compared to the common iterative exhaustive search algorithm, our strategy avoids the enumeration of all possible combinations in the universal set, reducing the time complexity from exponential to linear. Consequently, the constraint of time costs on the size of the universal set is greatly relaxed. Extensive evaluations of various super-resolution models show that DFSQ effectively improves performance even without fine-tuning. For example, for 4-bit EDSR×2 on the Urban benchmark, DFSQ obtains 0.242 dB PSNR gains.

关键词： super-resolution post-training quantization distribution-flexible subset quantization neural network

同方期刊数据库

Stability switches mechanism and PID control in reaction-diffusion network-organized neural systems with transmission delays

在线全文

学校读者我要写书评

暂无评论

Science China(Technological Sciences) 2025年第2期68卷 199-215页

作者： Xiangyu DU Min XIAO Wenwu YU Jiajin HE college of Automation&college of Artificial intelligence Nanjing University of Posts and TelecommunicationsNanjing 210023China college of Mathematics Southeast UniversityNanjing 211189China

Traditionally,the spatiotemporal dynamics of neural networks have been analyzed in the locally continuous *** this modeling approach is straightforward and practical,it fails to encapsulate real-world networks’intricate topological structures and ***,the mechanisms of stability switches induced by time delays and diffusion effects in networkorganized systems are still ***,effective dynamic optimization strategies for network-organized models remain to be *** this study,we develop a delayed reaction-diffusion neural network model on complex *** system incorporates the effects of inter-nodal diffusion coupling and exhibits profound architectural *** then pioneer the proportional-integral-derivative(PID)feedback control into the network-organized model to modulate dynamic *** linear stability analysis is conducted firstly,demonstrating that Turing patterns and Hopf bifurcation can be induced in the controlled neural network by varying diffusion coefficients and time ***,the bifurcation direction is deduced via the center manifold ***,a series of simulations are performed to validate the theoretical analysis and substantiate the efficacy of the PID control *** results exhibit that the PID feedback controller can flexibly regulate the dynamics of the network-organized systems and possesses excellent disturbance rejection capabilities.

关键词： complex network neural networks PID control turing instability hopf bifurcation time delays turing pattern

维普期刊数据库

Pushing one pair of labels apart each time in multi-label learning: from single positive to full labels

在线全文

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2025年

作者： Xiang LI Xinrui WANG Songcan CHEN MIIT Key Laboratory of Pattern Analysis and Machine intelligence College of Computer Science and Technology/College of Artificial Intelligence Nanjing University of Aeronautics and Astronautics

In multi-label learning (MLL), it is extremely challenging to accurately annotate every appearing object due to expensive costs and limited knowledge. When facing such a challenge, a more practical and cheaper alternative should be single positive multi-label learning (SPMLL), where only one positive label needs to be provided per sample. Existing SPMLL methods usually assume unknown labels as negatives, which inevitably introduces false negatives as noisy labels. More seriously, binary cross entropy (BCE) loss is often used for training, which is notoriously not robust to noisy labels. To mitigate this issue, we customize an objective function for SPMLL by pushing only one pair of labels apart each time to suppress the domination of negative labels, which is the main culprit of fitting noisy labels in SPMLL. To further combat such noisy labels, we explore the high-rankness of the label matrix, which can also push apart different labels. By directly extending from SPMLL to MLL with full labels, a unified loss applicable to both settings is derived. As a byproduct, the proposed loss can alleviate the imbalance inherent in MLL. Experiments on real datasets demonstrate that the proposed loss not only performs more robustly to noisy labels for SPMLL but also works well for full labels. Besides, we empirically discover that high-rankness can mitigate the dramatic performance drop in SPMLL. Most surprisingly, even without any regularization or fine-tuned label correction, only adopting our loss defeats state-of-the-art SPMLL methods on CUB, a dataset that severely lacks labels.

关键词：

Exit options sustain altruistic punishment and decrease the second-order free-riders, but it is not a panacea

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2025年第1期68卷 278-291页

作者： Chen SHEN Zhao SONG Lei SHI Jun TANIMOTO Zhen WANG Faculty of Engineering Sciences Kyushu University School of Statistics and Mathematics Yunnan University of Finance and Economics School of computing Engineering and Digital Technologies Teesside University School of Artificial intelligence OPtics and Electro Nics (iOPEN) Northwestern Polytechnical University

Why do humans punish free riders at their own expense? This behavior represents an evolutionary puzzle in human societies. This study explores the role of exit strategies in fostering altruistic punishment within evolutionary game theory. We extend the traditional prisoner's dilemma model by incorporating exiters, players who opt out for a small payoff while nullifying their opponent's payoff, and altruistic punishers who cooperate and punish non-cooperators. Our findings indicate that in well-mixed populations, exiters destabilize defection but do not promote altruistic punishment. In social networks, however,exiters enable altruistic punishment via cyclic dominance among altruistic punishers, defectors, and exiters. Notably, this cyclic dominance is sensitive to exit payoffs; adjustments can lead to cyclic dominance of non-punishing cooperators, defectors, and exiters, or to a bi-stable state between these two types of cyclic dominance. These results highlight the nuanced impact of exiters on altruistic punishment, emphasizing the need for careful incentivization of exit behavior. While exiters can support altruistic punishment in networked populations, their effectiveness is not a panacea and is highly sensitive to exit payoffs, indicating limits to the voluntary participation mechanism.

关键词： evolutionary game theory altruistic punishment coexistence cyclic dominance bi-stable

Distributed Economic Dispatch Algorithms of Microgrids Integrating Grid-Connected and Isolated Modes

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

IEEE/CAA Journal of Automatica Sinica 2025年第1期12卷 86-98页

作者： Zhongxin Liu Yanmeng Zhang Yalin Zhang Fuyong Wang IEEE the college of Artificial intelligence Nankai University and also with Tianjin Key Laboratory of Interventional BrainComputer Interface and Intelligent Rehabilitation Nankai University

The economic dispatch problem(EDP) of microgrids operating in both grid-connected and isolated modes within an energy internet framework is addressed in this paper. The multi-agent leader-following consensus algorithm is employed to address the EDP of microgrids in grid-connected mode, while the push-pull algorithm with a fixed step size is introduced for the isolated mode. The proposed algorithm of isolated mode is proven to converge to the optimum when the interaction digraph of microgrids is strongly connected. A unified algorithmic framework is proposed to handle the two modes of operation of microgrids simultaneously, enabling our algorithm to achieve optimal power allocation and maintain the balance between power supply and demand in any mode and any mode switching. Due to the push-pull structure of the algorithm and the use of fixed step size,the proposed algorithm can better handle the case of unbalanced graphs, and the convergence speed is improved. It is documented that when the transmission topology is strongly connected and there is bi-directional communication between the energy router and its neighbors, the proposed algorithm in composite mode achieves economic dispatch even with arbitrary mode ***, we demonstrate the effectiveness and superiority of our algorithm through numerical simulations.

关键词： Consensus algorithm distributed optimization economic dispatch (ED) energy router (ER) multi-agent systems

Improving Multi-task GNNs for Molecular Property Prediction via Missing Label Imputation

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Machine intelligence Research 2025年第1期22卷 131-144页

作者： Fenyu Hu Dingshuo Chen Qiang Liu Shu Wu University of Chinese Academy of Sciences Beijing100190China Center for Research on Intelligent Perception and computing State Key Laboratory of Multimodal Artificial Intelligence SystemsInstitute of AutomationChinese Academy of SciencesBeijing100190China

The prediction of molecular properties is a fundamental task in the field of drug ***,graph neural networks(GNNs)have been gaining prominence in this *** a molecule tends to have multiple correlated properties,there is a great need to develop the multi-task learning ability of ***,limited by expensive and time-consuming human annotations,collecting complete labels for each task is *** a result,most existing benchmarks involve many missing labels in training data,and the performance of GNNs is impaired due to the lack of sufficient supervision *** overcome this obstacle,we propose to improve multi-task molecular property prediction by missing label ***,a bipartite graph is first introduced to model the molecule-task co-occurrence ***,the imputation of missing labels is transformed into predicting missing edges on this bipartite *** predict the missing edges,a graph neural network is devised,which can learn the complex molecule-task co-occurrence *** that,we select reliable pseudo labels according to the uncertainty of the prediction *** with enough and reliable supervision information,our approach achieves state-of-the-art performance on a variety of real-world datasets.

关键词： Graph classification imbalance learning prediction bias mixture of experts multiview representations

Node ranking based on graph curvature and PageRank

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Chinese Physics B 2025年第2期34卷 496-507页

作者： Hongbo Qu Yu-Rong Song Ruqi Li Min Li Guo-Ping Jiang School of Computer Science Nanjing University of Posts and TelecommunicationsNanjing 210023China college of Automation&college of Artificial intelligence Nanjing University of Posts and TelecommunicationsNanjing 210023China

Identifying key nodes in complex networks is crucial for understanding and controlling their dynamics. Traditional centrality measures often fall short in capturing the multifaceted roles of nodes within these networks. The Page Rank algorithm, widely recognized for ranking web pages, offers a more nuanced approach by considering the importance of connected nodes. However, existing methods generally overlook the geometric properties of networks, which can provide additional insights into their structure and functionality. In this paper, we propose a novel method named Curv-Page Rank(C-PR), which integrates network curvature and Page Rank to identify influential nodes in complex networks. By leveraging the geometric insights provided by curvature alongside structural properties, C-PR offers a more comprehensive measure of a node's influence. Our approach is particularly effective in networks with community structures, where it excels at pinpointing bridge nodes critical for maintaining connectivity and facilitating information flow. We validate the effectiveness of C-PR through extensive experiments. The results demonstrate that C-PR outperforms traditional centrality-based and Page Rank methods in identifying critical nodes. Our findings offer fresh insights into the structural importance of nodes across diverse network configurations, highlighting the potential of incorporating geometric properties into network analysis.

关键词： important nodes graph curvature complex networks network geometry

Attar: RRAM-based in-memory attention accelerator with software-hardware co-optimization

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2025年第03期68卷 371-387页

作者： Bing LI Ying QI Ying WANG Yinhe HAN Information Engineering college Capital Normal University Research Center for Intelligent computing Systems Institute of Computing TechnologyChinese Academy of Sciences

The attention mechanism has become a pivotal component in artificial intelligence, significantly enhancing the performance of deep learning applications. However, its quadratic computational complexity and intricate computations lead to substantial inefficiencies when processing long sequences. To address these challenges, we introduce Attar, a resistive random access memory(RRAM)-based in-memory accelerator designed to optimize attention mechanisms through software-hardware co-optimization. Attar leverages efficient Top-k pruning and quantization strategies to exploit the sparsity and redundancy of attention matrices, and incorporates an RRAM-based in-memory softmax engine by harnessing the versatility of the RRAM crossbar. Comprehensive evaluations demonstrate that Attar achieves a performance improvement of up to 4.88× and energy saving of 55.38% over previous computing-in-memory(CIM)-based accelerators across various models and datasets while maintaining comparable accuracy. This work underscores the potential of in-memory computing to enhance the efficiency of attention-based models without compromising their effectiveness.

关键词： RRAM computing-in-memory attention pruning quantization

同方期刊数据库