检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Wang, Yaxian Ding, Henghui He, Shuting Jiang, Xudong Wei, Bifan Liu, Jun School of Computer Science and Technology Xi'an Jiaotong University China Ministry of Education Key Laboratory of Intelligent Networks and Network Security Xi'an Jiaotong University China Institute of Big Data Fudan University China Shanghai University of Finance and Economics China Nanyang Technological University Singapore School of Continuing Education Xi'an Jiaotong University China Shaanxi Province Key Laboratory of Big Data Knowledge Engineering Xi'an Jiaotong University China

In this work, we address the challenging task of Generalized Referring Expression Comprehension (GREC). Compared to the classic Referring Expression Comprehension (REC) that focuses on single-target expressions, GREC extends the scope to a more practical setting by further encompassing notarget and multi-target expressions. Existing REC methods face challenges in handling the complex cases encountered in GREC, primarily due to their fixed output and limitations in multi-modal representations. To address these issues, we propose a Hierarchical Alignment-enhanced Adaptive Grounding Network (HieA2G) for GREC, which can flexibly deal with various types of referring expressions. First, a Hierarchical Multi-modal Semantic Alignment (HMSA) module is proposed to incorporate three levels of alignments, including word-object, phrase-object, and text-image alignment. It enables hierarchical cross-modal interactions across multiple levels to achieve comprehensive and robust multi-modal understanding, greatly enhancing grounding ability for complex cases. Then, to address the varying number of target objects in GREC, we introduce an Adaptive Grounding Counter (AGC) to dynamically determine the number of output targets. Additionally, an auxiliary contrastive loss is employed in AGC to enhance object-counting ability by pulling in multi-modal features with the same counting and pushing away those with different counting. Extensive experimental results show that HieA2G achieves new state-of-the-art performance on the challenging GREC task and also the other 4 tasks, including REC, Phrase Grounding, Referring Expression Segmentation (RES), and Generalized Referring Expression Segmentation (GRES), demonstrating the remarkable superiority and generalizability of the proposed HieA2G. Copyright © 2025, The Authors. All rights reserved.

关键词： Electric grounding

来源：评论

学校读者我要写书评

暂无评论

EncryptFlow: Efficient and Lossless Image Encryption Network Based on Normalizing Flows

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2025年

作者： Yang, Menglin Xie, Dong Zhang, Guiting Chen, Fulong Wang, Taochun Hu, Peng Anhui Normal University Anhui Provincial Key Laboratory of Industrial Intelligence Data Security Anhui Engineering Research Centers of Medical Big Data Intelligent System Wuhu241002 China Anhui Normal University School of Computer and Information Wuhu241002 China

Compared to cryptographic image encryption schemes, neural networks (NN) based image encryption schemes exhibit a significantly larger key space and offer enhanced capabilities for parallel processing of image data. However, most existing NN-based image encryption schemes suffer from high time complexity in generating random keys, and their decryption processes often fail to fully recover the plaintext images without loss. In this paper, we first propose a normalizing flows based encryption network, called EncryptFlow, designed to achieve efficient and lossless image encryption. Normalizing flows employ a special coupling structure to couple the partitioned data, thereby establishing interdependence among them. Specifically, we utilize coupling structures (e.g. additive coupling) that allows the image blocks to alternately encrypt each other during forward propagation. Additionally, we devise a key generation algorithm that produces sub-keys tailored for each layer of the encryption network. The proposed EncryptFlow network seamlessly integrates both encryption and decryption functionalities, leveraging the XOR operation as the encryption function within each layer. The experimental results and comparative analyses indicate that EncryptFlow can encrypt 256 × 256 grayscale images with an average time of merely 0.047s, and similarly, it requires only 0.188s to encrypt color images of the same dimensions. © 2020 IEEE.

关键词： Optical flows

来源：评论

学校读者我要写书评

暂无评论

A comprehensive survey on shadow removal from document images: datasets, methods, and opportunities

引用

Vicinagearth 2025年第1期2卷 1-18页

作者： Wang, Bingshu Li, Changping Zou, Wenbin Zhang, Yongjun Chen, Xuhang Chen, C.L. Philip School of Software Northwestern Polytechnical University Xi’an China Guangdong Provincial Key Laboratory of Intelligent Information Processing & Shenzhen Key Laboratory of Media Security Shenzhen University Shenzhen China Guangdong Key Laboratory of Intelligent Information Processing College of Electronics and Information Engineering Shenzhen University Shenzhen China Yongjun Zhang is with the State Key Laboratory of Public Big Data College of Computer Science and Technology Guizhou University Guiyang China School of Computer Science and Engineering Huizhou University Huizhou China School of Computer Science and Engineering South China University of Technology and Pazhou Lab Guangzhou China

With the rapid development of document digitization, people have become accustomed to capturing and processing documents using electronic devices such as smartphones. However, the captured document images often suffer from issues like shadows and noise due to environmental factors, which can affect their readability. To improve the quality of captured document images, researchers have proposed a series of models or frameworks and applied them in distinct scenarios such as image enhancement, and document information extraction. In this paper, we primarily focus on shadow removal methods and open-source datasets. We concentrate on recent advancements in this area, first organizing and analyzing nine available datasets. Then, the methods are categorized into conventional methods and neural network-based methods. Conventional methods use manually designed features and include shadow map-based approaches and illumination-based approaches. Neural network-based methods automatically generate features from data and are divided into single-stage approaches and multi-stage approaches. We detail representative algorithms and briefly describe some typical techniques. Finally, we analyze and discuss experimental results, identifying the limitations of datasets and methods. Future research directions are discussed, and nine suggestions for shadow removal from document images are proposed. To our knowledge, this is the first survey of shadow removal methods and related datasets from document images.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Towards a Smaller Student: Capacity Dynamic Distillation for Efficient Image Retrieval

arXiv

引用

arXiv 2023年

作者： Xie, Yi Zhang, Huaidong Xu, Xuemiao Zhu, Jianqing He, Shengfeng South China University of Technology China Huaqiao University China Singapore Management University Singapore State Key Laboratory of Subtropical Building Science China Ministry of Education Key Laboratory of Big Data and Intelligent Robot China Guangdong Provincial Key Lab of Computational Intelligence and Cyberspace Information China

Previous Knowledge Distillation based efficient image retrieval methods employs a lightweight network as the student model for fast inference. However, the lightweight student model lacks adequate representation capacity for effective knowledge imitation during the most critical early training period, causing final performance degeneration. To tackle this issue, we propose a Capacity Dynamic Distillation framework, which constructs a student model with editable representation capacity. Specifically, the employed student model is initially a heavy model to fruitfully learn distilled knowledge in the early training epochs, and the student model is gradually compressed during the training. To dynamically adjust the model capacity, our dynamic framework inserts a learnable convolutional layer within each residual block in the student model as the channel importance indicator. The indicator is optimized simultaneously by the image retrieval loss and the compression loss, and a retrieval-guided gradient resetting mechanism is proposed to release the gradient conflict. Extensive experiments show that our method has superior inference speed and accuracy, e.g., on the VeRi-776 dataset, given the ResNet101 as a teacher, our method saves 67.13% model parameters and 65.67% FLOPs without sacrificing accuracy. Copyright © 2023, The Authors. All rights reserved.

关键词： Image retrieval

来源：评论

学校读者我要写书评

暂无评论

Graph Structure Prefix Injection Transformer for Multi-Modal Entity Alignment

SSRN

引用

SSRN 2024年

作者： Li, Zhifei Luo, Xiangyu Zhang, Miao Zhang, Yan Xiao, Kui School of Computer Science and Information Engineering Hubei University Wuhan430062 China School of Cyber Science and Technology Hubei University Wuhan430062 China Key Laboratory of Intelligent Sensing System and Security Hubei University Ministry of Education Wuhan430062 China Hubei Key Laboratory of Big Data Intelligent Analysis and Application Hubei University Wuhan430062 China

Multi-modal entity alignment aims to integrate equivalent entities in diverse multi-modal knowledge graphs. However, previous studies have primarily focused on multi-modal fusion methods without considering the impact of graph structure heterogeneity on entity alignment tasks due to the distinct structural features of different knowledge graphs. Moreover, the multi-modal information that enriches entity representations, including image, textual content, and numerical attributes, frequently incorporates noise due to modal discrepancies. To tackle these challenges, we introduce the Graph Structural Prefix Injection transformer (GSIEA) for multi-modal entity alignment. This novel approach minimizes structural heterogeneity’s impact and optimizes structural data utilization by integrating a structural prefix inject interaction module within the multi-head attention mechanism. Then, we adopt a convolutional enhancement module to extract local features of multi-modal information and complete feature fusion by cross-modal weights. Extensive experiments on two datasets demonstrate that GSIEA outperforms previous baseline models, showcasing an average performance improvement of 3.26% on MRR, with a maximum enhancement of 12.5%. Additionally, the average performance improvement on Hits@1 is 4.96%, with a maximum enhancement of 16.92%. Our code is available at https://***/HubuKG/GSIEA. © 2024, The Authors. All rights reserved.

关键词： Knowledge graph

来源：评论

学校读者我要写书评

暂无评论

Large Language Models Enhanced Personalized Graph Neural Architecture Search in Federated Learning 39

Large Language Models Enhanced Personalized Graph Neural Arc...

引用

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Fang, Hui Gao, Yang Zhang, Peng Yao, Jiangchao Chen, Hongyang Wang, Haishuai Zhejiang Key Laboratory of Accessible Perception and Intelligent Systems College of Computer Science Zhejiang University China Cyberspace Institute of Advanced Technology Guangzhou University China Cooperative Medianet Innovation Center Shanghai Jiaotong University China Research Center for Data Hub and Security Zhejiang Lab China

ISBN: (纸本)157735897X

Personalized federated learning (PFL) on graphs is an emerging field focusing on the collaborative development of architectures across multiple clients, each with distinct graph data distributions while adhering to strict privacy standards. This area often requires extensive expert intervention in model design, which is a significant limitation. Recent advancements have aimed to automate the search for graph neural network architectures, incorporating large language models (LLMs) for their advanced reasoning and self-reflection capabilities. However, two technical challenges persist. First, although LLMs are effective in natural language processing, their ability to meet the complex demands of graph neural architecture search (GNAS) is still being explored. Second, while LLMs can guide the architecture search process, they do not directly solve the issue of client drift due to heterogeneous data distributions. To address these challenges, we introduce a novel method, Personalized Federated Graph Neural Architecture Search (PFGNAS). This approach employs a task-specific prompt to identify and integrate optimal GNN architectures continuously. To counteract client drift, PFGNAS utilizes a weight-sharing strategy of supernet, which optimizes the local architectures while ensuring client-specific personalization. Extensive evaluations show that PFGNAS significantly outperforms traditional PFL methods, highlighting the advantages of integrating LLMs into personalized federated learning environments. Copyright © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

Point Cloud Compression Method Based on Curvature Grading and Inter-point Compression for Industrial Products

Point Cloud Compression Method Based on Curvature Grading an...

引用

International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery, CyberC

作者： Lei Wang Wei Huang Yiping Chen Ran Chen School of Computer Nanjing University of Posts and Telecommunications Nanjing P. R. China Jiangsu Key Laboratory of Big Data Security and Intelligent Processing Nanjing P. R. China

The traditional point cloud compression methods fail to meet the compression requirements of complex surface point cloud files in the industrial field. This paper proposes a point cloud compression method suitable for industrial products with defective and complex surfaces, which compresses the defect and non-defect parts separately using different compression approaches. A curvature-based compression method is designed for the defect points, and an inter-point compression method is developed for the non-defect part, achieving compression of both components of the point cloud. The experimental results show the advantages of the proposed compression method through a comparative analysis of time efficiency, compression ratio, and reconstruction accuracy for industrial point cloud compression.

关键词：

来源：评论

学校读者我要写书评

暂无评论

GNEA: A Novel GCN-based Network Embedding Algorithm for Semantic Social Network

GNEA: A Novel GCN-based Network Embedding Algorithm for Sema...

引用

International Symposium on System and Software Reliability (ISSSR)

作者： Jinbo Zhao Jinling He Yuanhan Du Ming Tang Xiaolong Xu Jiangsu Key Laboratory of Big Data Security & Intelligent Processing Nanjing University of Posts and Telecommunications Nanjing China State Grid Jiangsu Electric Power Company Limited Nanjing China School of Computer Science Nanjing University of Posts and Telecommunications Nanjing China

ISBN: (数字)9798350362930

ISBN: (纸本)9798350362947

With the development of semantic social networks, social networks become more complex and their size expands rapidly, which brings significant challenges to social network analysis. Network Embedding can transform the network from a high-dimensional adjacency matrix to a low-dimensional one, which has excellent foreground in link prediction, vertex classification, and graph visualization. Nowadays, most algorithms are proposed for static networks with little consideration of the dynamic changes in the network. However, most real-world networks change over time. Directly applying static network embedding algorithms to dynamic networks will result in poor stability, flexibility, and efficiency. Because of these shortcomings, we propose a graph convolution-based dynamic network representation learning algorithm (GNEA), which combines Graph Convolutional Network (GCN) with Long Short-term Memory (LSTM) to train the time series networks. GCN is applied to extract features in the network, and LSTM is utilized to capture temporal information. GNEA can not only perform characterization learning on the network but also obtain more abundant information during the dynamic evolution of the network. GNEA is compared with several typical algorithms on the datasets Stochastic Block Model (SBM), Autonomous Systems (AS), and Digital Bibliography and Library Project (DBLP). The results prove that GNEA has good performance in MAP and precision@k.

关键词： Representation learning Accuracy Social networking (online) Heuristic algorithms Semantics Prediction algorithms Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Dlr: Adversarial Examples Detection and Label Recovery for Deep Neural Networks

SSRN

引用

SSRN 2024年

作者： Han, Keji Li, Yun Ge, Yao Wang, Ruchuan Nanjing University of Posts and Telecommunications Wenyuan Road 9 Nanjing210046 China Jiangsu Key Laboratory of Big Data Security & Intelligent Processing Wenyuan Road 9 Nanjing210046 China Jiangsu High Technology Research Key Laboratory for Wireless Sensor Networks Wenyuan Road 9 Nanjing210046 China

Deep neural networks (DNNs) are demonstrated to be vulnerable to the adversarial example crafted by the adversary to fool the target model. Adversarial training and adversarial example detection are two popular methods to mitigate the adversarial issue. The adversarial training makes the target model recognize the adversarial example correctly for the image classification task, while it does not generalize well. The adversarial detection generalizes well but cannot help the target model recognize the adversarial example. We first define the label recovery task to address the adversarial issue for DNNs. Then we propose a new generative classifier and apply it to the adversarial example label recovery task. The proposed method is named as Detection and Label Recovery (DLR) defense method, which consists of Detector and Recover. Detector feeds the legitimate and adversarial examples to the target model and Recover, respectively. The Recover explores the ground-truth label of the detected adversarial example. DLR combines the merits of adversarial training and adversarial example detection. Experimental results demonstrate that our method outperforms some state-of-the-art adversarial training methods. © 2024, The Authors. All rights reserved.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Path Planning based on Reinforcement Learning with Improved APF model for Synergistic Multi-UAVs

Path Planning based on Reinforcement Learning with Improved ...

引用

International Conference on Computer Supported Cooperative Work in Design

作者： Qun Ding Xiaolong Xu Wenming Gui Jiangsu Key Laboratory of Big Data Security & Intelligent Processing Nanjing China School of Computer Science Nanjing University of Posts and Telecommunications Nanjing China School of Computer Science Jinling Institute of Technology Nanjing China

As the emerging technology of Unmanned Aerial Vehicles (UAVs) becomes mature, UAVs are widely used in environmental monitoring, communication and other fields. In view of this, this paper analyzed the task of synergistic multi-UAVs exploration of unknown environments, and proposed a path planning method for them based on reinforcement learning. Firstly, the path planning task of the UAVs was divided into two parts: the path travel strategy module and the information exploration strategy module. Models of the two modules were based on the Deep Deterministic Policy Gradient algorithm (DDPG), and an improved Artificial Potential Field (APF) force traction mechanism was introduced in the path travel strategy module. Its aim was to assist in guiding the generation of UAV flight path trajectories. Also it could enhance the learning capability of the model. The path travel strategy module would generate the complete flight path of the whole cluster in a distributed manner. A series of temporary target points provided by the information exploration strategy module helped. In maps with 21.5%, 25.3% and 29.6% of obstacles, multi-UAVs could achieve 84.2%, 76.7% and 69.9% of environmental exploration by the designed method. Compared with the APF method, the A star method and the Breath First Search (BFS) method, the proposed method is not only able to plan feasible paths in a more complex map model, but also the curvature of the planned paths is smoother, thus achieving the goal of reducing the energy cost of UAVs.

关键词：

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：