检索结果-内蒙古大学图书馆

Global video object segmentation with spatial constraint module

Computational Visual Media 2023年第2期9卷 385-400页

作者： Yadang Chen Duolin Wang Zhiguo Chen Zhi-Xin Yang Enhua Wu Engineering Research Center of Digital Forensics Ministry of EducationSchool of Computer and SoftwareNanjing University of Information Science and TechnologyNanjing 210044China State Key Laboratory of Internet of Things for Smart City Department of Electromechanical EngineeringUniversity of MacaoMacao 999078China State Key Laboratory of Computer Science Institute of SoftwareUniversity of Chinese Academy of SciencesBeijing 100190China Faculty of Science and Technology University of MacaoMacao 999078China

We present a lightweight and efficient semisupervised video object segmentation network based on the space-time memory *** some extent,our method solves the two difficulties encountered in traditional video object segmentation:one is that the single frame calculation time is too long,and the other is that the current frame’s segmentation should use more information from past *** algorithm uses a global context(GC)module to achieve highperformance,real-time *** GC module can effectively integrate multi-frame image information without increased memory and can process each frame in real ***,the prediction mask of the previous frame is helpful for the segmentation of the current frame,so we input it into a spatial constraint module(SCM),which constrains the areas of segments in the current *** SCM effectively alleviates mismatching of similar targets yet consumes few additional *** added a refinement module to the decoder to improve boundary *** model achieves state-of-the-art results on various datasets,scoring 80.1%on YouTube-VOS 2018 and a J&F score of 78.0%on DAVIS 2017,while taking 0.05 s per frame on the DAVIS 2016 validation dataset.

关键词： video object segmentation semantic segmentation global context(GC)module spatial constraint

来源：评论

学校读者我要写书评

暂无评论

Resisting Over-Smoothing in Graph Neural Networks via Dual-Dimensional Decoupling 24

Resisting Over-Smoothing in Graph Neural Networks via Dual-D...

引用

32nd ACM International Conference on Multimedia, MM 2024

作者： Shen, Wei Ye, Mang Huang, Wenke National Engineering Research Center for Multimedia Software School of Computer Science Wuhan University Wuhan China

ISBN: (纸本)9798400706868

Graph Neural Networks (GNNs) are widely employed to derive meaningful node representations from graphs. Despite their success, deep GNNs frequently grapple with the oversmoothing issue, where node representations become highly indistinguishable due to repeated aggregations. In this work, we consider the oversmoothing issue from two aspects of the node embedding space: dimension and instance. Specifically, while existing methods primarily concentrate on instance-level node relations to mitigate oversmoothing, we propose to mitigate oversmoothing at dimension level. We reveal the heightened information redundancy between dimensions which diminishes information diversity and impairs node differentiation in GNNs. Motivated by this insight, we propose the Dimension-Level Decoupling (DLD) to reduce dimension redundancy, enhancing dimensional-level node differentiation. Besides, at the instance level, the neglect of class differences leads to vague classification boundaries. Hence, we introduce the Instance-Level Class-Difference Decoupling (ICDD) that repels inter-class nodes and attracts intra-class nodes, improving the instance-level node discrimination with clear classification boundaries. Additionally, we introduce a novel evaluation metric that considers the impact of class differences on node distances, facilitating precise oversmoothing measurement. Extensive experiments demonstrate the effectiveness of our method Dual-Dimensional Class-Difference Decoupling (DDCD) across diverse scenarios. © 2024 ACM.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

Anomaly Detection in Spatial Information Networks: A Security Knowledge Graph-based Approach

Anomaly Detection in Spatial Information Networks: A Securit...

引用

2024 International Conference on Intelligent Computing and Robotics, ICICR 2024

作者： Tan, Yayue Zhang, Yaqi Liu, Hui Tang, Yongqing College of Computer and Information Science School of Software Southwest University Chongqing China School of Information and Software Engineering University of Electronic Science and Technology of China Chengdu China

ISBN: (纸本)9798350372274

Anomaly detection stands as a critical element in securing space information networks (SINs). This paper delves into the realm of anomaly detection within dynamic networks, shedding light on established methodologies. The focus lies on scrutinizing various anomaly detection strategies in space information networks, conducting a comprehensive assessment of their practical applications, advantages, and drawbacks. Moreover, it introduces a novel anomaly detection approach driven by deep learning principles. This innovative method merges graph convolution techniques with a network securityoriented knowledge graph, presenting a cutting-edge solution. As a final contribution, the paper extends its purview to the future landscape of anomaly detection within space information networks, offering valuable insights into potential advancements and forthcoming trends in this domain. This holistic investigation seeks to enhance the understanding of anomaly detection paradigms within the unique context of space information networks, aiming to fortify their security measures and operational resilience. ©2024 IEEE.

关键词： Knowledge graph

来源：评论

学校读者我要写书评

暂无评论

Dual-Enhanced High-Order Self-Learning Tensor Singular Value Decomposition for Robust Principal Component Analysis

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2024年第7期5卷 3564-3578页

作者： Xu, Honghui Fang, Chuangjie Wang, Renfang Chen, Shengyong Zheng, Jianwei Zhejiang University of Technology College of Computer Science and Technology Hangzhou310023 China Zhejiang Wanli University College of Big Data and Software Engineering Ningbo315200 China Tianjin University of Technology College of Computer Sciences and Engineering Tianjin300384 China

Recently, tensor singular value decomposition (TSVD) within high-order (Ho) algebra framework has shed new light on tensor robust principal component analysis (TRPCA) problem. However, HoTSVD lacks flexibility in handling the hidden correlations along different modes of large-scale multidimensional data. Moreover, the utilization of fixed or data-independent transformations in HoTSVD may result in suboptimality. For a relief, we propose a dual-enhanced self-learning TSVD along all modes to address computational flaws and learn a lossless transformation that induces a lower average-rank tensor. Specifically, we multiply the learnable semiorthogonal matrices obtained through Tucker compression with the original tensor along all modes, thus obtaining a core tensor with more inherent low rankness. Building upon this foundation, a new TNN is introduced by generalizing HoTSVD to Mode-k TSVD, followed by the facilitation to the core tensor, achieving dual-enhancement. Moreover, a reweighting scheme is imposed on the Mode-k HoTSVD to learn the global low-rank correlation and provide an efficient numerical solution. Finally, an alternating direction method of multipliers (ADMM)-based algorithm is developed as a solver. Experimental results on several types of multidimensional visual data, including light field images (LFI) and color videos, demonstrate the superiority of the proposal over previous state-of-the-art methods. © 2020 IEEE.

关键词： Discrete Fourier transforms

来源：评论

学校读者我要写书评

暂无评论

An Approach to Optimize Symbolic Execution in Ethereum Smart Contracts 24

An Approach to Optimize Symbolic Execution in Ethereum Smart...

引用

24th IEEE International Conference on software Quality, Reliability and Security Companion, QRS-C 2024

作者： Liu, Yi Cai, Lizhi Chai, Chenchen Ma, Yuan School of Information Science and Engineering East China University of Science and Technology Shanghai China Shanghai Development Center of Computer Software Technology Shanghai Key Laboratory of Computer Software Testing & Evaluating Shanghai China

ISBN: (纸本)9798350365658

As the application of smart contracts in blockchain technology becomes increasingly widespread, their security issues have emerged as a focal point of both research and practice. Although symbolic execution technology has been extensively applied for the detection of security vulnerabilities in smart contracts, its limitations in terms of execution efficiency and code coverage remain significant. In response to these challenges, this study introduces an innovative method named HoneyToken-DetectorRev, which optimizes traditional symbolic execution through program slicing technology, aiming to enhance the efficiency and effectiveness of smart contract security detection. By selecting ten smart contracts from the Ethereum platform, which vary in business logic and complexity, we conducted an empirical analysis of HoneyToken-DetectorRev. The experimental results demonstrate that this method significantly improves the efficiency of smart contract detection without sacrificing precision. This research not only provides an effective technical approach for the security detection of smart contracts but also offers theoretical and practical guidance for further enhancing the security of blockchain applications. © 2024 IEEE.

关键词： Smart contract

来源：评论

学校读者我要写书评

暂无评论

TrafficNet: A Hybrid CNN-FNN Model for Analysis of Traffic Accidents in Seoul

引用

Journal of Computing science and engineering 2023年第4期17卷 182-194页

作者： Mustafa, Ghulam Cho, Seong-je Hwang, Youngsup Department of Computer Engineering Sun Moon University Asan Korea Republic of Department of Software Science Dankook University Yongin Korea Republic of

The escalating global trend of traffic accidents with subsequent loss of lives is a matter of grave concern that requires immediate attention. Extensive efforts have been made to mitigate accidents and develop effective prevention strategies. This research paper focuses on a comprehensive analysis of traffic accidents in Seoul, aiming to identify factors and accident types that contribute to increased severity. To achieve this, we introduced a new approach called "TrafficNet: A Hybrid CNN-FNN Model" to evaluate effects of various parameters on the severity of traffic accidents in Seoul. Our main objective was to classify accidents into four distinct levels of severity: minor injuries, slander, fatalities, and injury reports. To assess the effectiveness of our proposed model, we conducted comprehensive experiments using publicly available traffic accident data provided by Seoul Metropolitan Government. These experiments involved six different models, including five machine learning models (decision tree, random forest, k-nearest neighbor, gradient boosting, and support vector machine) and one deep learning model (multilayer perceptron). The proposed model demonstrated exceptional performance, surpassing all other models and previous research findings using the same dataset. On the test dataset, TrafficNet achieved an impressive accuracy of 93.98% with a precision of 94.31%, a recall of 93.98%, and an F1-score of 93.89%. Copyright 2023. The Korean Institute of Information Scientists and Engineers

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Reconstruct and Match: Out-of-Distribution Robustness via Topological Homogeneity 38

Reconstruct and Match: Out-of-Distribution Robustness via To...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Chen, Chaoqi Tang, Luyao Huang, Hui College of Computer Science and Software Engineering Shenzhen University China School of Informatics Xiamen University China

Since deep learning models are usually deployed in non-stationary environments, it is imperative to improve their robustness to out-of-distribution (OOD) data. A common approach to mitigate distribution shift is to regularize internal representations or predictors learned from in-distribution (ID) data to be domain invariant. Past studies have primarily learned pairwise invariances, ignoring the intrinsic structure and high-order dependencies of the data. Unlike machines, humans recognize objects by first dividing them into major components and then identifying the topological relation of these components. Motivated by this, we propose Reconstruct and Match (REMA), a general learning framework for object recognition tasks to endow deep models with the capability of capturing the topological homogeneity of objects without human prior knowledge or fine-grained annotations. To identify major components from objects, REMA introduces a selective slot-based reconstruction module to dynamically map dense pixels into a sparse and discrete set of slot vectors in an unsupervised manner. Then, to model high-order dependencies among these components, we propose a hypergraph-based relational reasoning module that models the intricate relations of nodes (slots) with structural constraints. Experiments on standard benchmarks show that REMA outperforms state-of-the-art methods in OOD generalization and test-time adaptation settings. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

SeerahGPT: Retrieval Augmented Generation based Large Language Model 18

SeerahGPT: Retrieval Augmented Generation based Large Langua...

引用

18th International Conference on Open Source Systems and Technologies, ICOSST 2024

作者： Obaid, Surayya Bawany, Narmeen Zakaria Center for Computing Research Department of Computer Science and Software Engineering Jinnah Univesity for Women Karachi Pakistan

ISBN: (纸本)9798331508692

Large Language Models(LLMs) have become widely recognized in recent years for their exceptional performance in language generation capabilities. As a result, an unprecedented rise is seen in its use cases in various domains specifically involving Natural Language Processing(NLP). These models however perform suboptimally when exploited in the field of studies where authenticity of the generated content is a critical aspect. One such domain is usage of LLM for exploration of the life of Prophet Muhammad S.A.W(commonly referred to as Seerah). It is of utmost significance to ensure the authenticity and reliability in the sources used and reported by the LLM due to the sensitive nature of the domain. The contemporary LLMs, however, lack the explainability in their response due to their inherent black-box nature. In our study, we have presented a novel LLM named SeerahGPT that addresses this challenge with the help of retrieval-augmented generation (RAG). This technique enables the model to utilize both parametric and nonparametric memories for generating response of queries. Our model, built on the Llama-2-7b architecture, employs Sentence Transformer embedding to effectively retrieve relevant information. The model's capabilities are augmented by integrating it with a corpus having Islamic texts such as the Quranic translation and Hadith collections, and historical accounts. The model's performance is benchmarked against its base model using both quantitative and qualitative metrics. The comparative analysis with Llama-2-7b revealed that SeerahGPT incorporation with external knowledge sources, provided more authentic and verifiable responses, despite the others exhibiting greater fluency. Performance metrics such as BLEU, ROUGE, and METEOR indicated SeerahGPT's better accuracy and contextual handling. This study paves way for analysis of such sensitive domains in more efficient way that can be utilized in other complex domains such as Islamic theology and Fiqh or legal

关键词： Natural language processing systems

来源：评论

学校读者我要写书评

暂无评论

Cutting-Edge AI-Driven Higher Educational and Career Advisory Platform

Cutting-Edge AI-Driven Higher Educational and Career Advisor...

引用

2024 International Conference on Innovative Computing, Intelligent Communication and Smart Electrical Systems, ICSES 2024

作者： Dewasurendra, S.V. Udunuwara, U.K.C. Wanniarachchi, T.T. Panditharathne, P.D.R.L. Thelijjagoda, Samantha Dissanayaka, Kapila Sri Lanka Institute of Information Technology Department of Computer Science and Software Engineering Malabe Sri Lanka

ISBN: (纸本)9798331543617

Sri Lankan students, especially those from rural and low-income backgrounds, face significant challenges in accessing higher education, including university applications, career planning, and financial aid. This research introduces an AI-powered platform designed to provide personalized assistance through university course recommendations, career guidance, financial aid information, and AI-generated mock aptitude tests. The platform employs Graph Neural Networks (GNNs) for course recommendations, Recurrent Neural Networks (RNNs) for career guidance, and a Retrieval-Augmented Generation (RAG) model for financial aid navigation. Additionally, AI-driven algorithms with Optical Character Recognition (OCR) technology support exam preparation. Data from students and universities informed model training, and evaluations showed high accuracy and user satisfaction across components. Results indicated that the platform effectively aligns educational resources with students' academic and career goals, promoting greater educational accessibility and support for underserved Sri Lankan students. © 2024 IEEE.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

Protein-Protein Interaction Sites Prediction Using Graph Convolutional Networks

Protein-Protein Interaction Sites Prediction Using Graph Con...

引用

2024 International Conference on computer and Applications, ICCA 2024

作者： Alkhateeb, Noor Jamal Awad, Mamoun United Arab Emriates University Department of Computer Science and Software Engineering Alain United Arab Emirates

ISBN: (纸本)9798350367560

Protein-protein interactions (PPI) are essential in keeping the cells functioning properly. Identifying PPI binding sites is a fundamental problem in system Biology, and it contributes to a better understanding of lower organisms such as viruses, a multitude-drug design and vaccines. The experimental methods for PPI binding sites identification are slow and expensive. Therefore, great research efforts have been attempted to improve the performance of computational methods. In this paper, we use a deep learning model based on Graph Convolutional Networks (GCN) to predict putative interaction sites on the surface of an isolated protein. We extracted features from both protein sequence and structure to enhance the accuracy of PPI binding sites predictions. Our model achieved higher accuracy compared to other models. © 2024 IEEE.

关键词： Convolutional neural networks

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：