检索结果-内蒙古大学图书馆

International Conference on Computer Vision (ICCV)

作者： Weiying Zheng Cheng Xu Xuemiao Xu Wenxi Liu Shengfeng He South China University of Technology State Key Laboratory of Subtropical Building Science Ministry of Education Key Laboratory of Big Data and Intelligent Robot Guangdong Provincial Key Lab of Computational Intelligence and Cyberspace Information Fuzhou University Singapore Management University

Video inpainting aims at filling in missing regions of a video. However, when dealing with dynamic scenes with camera or object movements, annotating the inpainting target becomes laborious and impractical. In this paper, we resolve the one-shot video inpainting problem in which only one annotated first frame is provided. A naive solution is to propagate the initial target to the other frames with techniques like object tracking. In this context, the main obstacles are the unreliable propagation and the partially inpainted artifacts due to the inaccurate mask. For the former problem, we propose curricular inactivation to replace the hard masking mechanism for indicating the in-painting target, which is robust to erroneous predictions in long-term video inpainting. For the latter, we explore the properties of inpainting residue and present an online residue removal method in an iterative detect-and-refine manner. Extensive experiments on several real-world datasets demonstrate the quantitative and qualitative superiorities of our proposed method in one-shot video inpainting. More importantly, our method is extremely flexible that can be integrated with arbitrary traditional inpainting models, activating them to perform the reliable one-shot video inpainting task. Video demonstrations can be found in our supplement, and our code can be found at https://***/Arise-zwy/CIRI.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Rotation-Adaptive Point Cloud Domain Generalization via Intricate Orientation Learning

arXiv

引用

arXiv 2025年

作者： Liu, Bangzhen Zheng, Chenxi Xu, Xuemiao Xu, Cheng Zhang, Huaidong He, Shengfeng South China University of Technology Guangzhou China Guangdong Engineering Center for Large Model and GenAI Technology The State Key Laboratory of Subtropical Building and Urban Science The Ministry of Education Key Laboratory of Big Data and Intelligent Robot The Guangdong Provincial Key Lab of Computational Intelligence and Cyberspace Information China Singapore Management University Singapore

The vulnerability of 3D point cloud analysis to unpredictable rotations poses an open yet challenging problem: orientation-aware 3D domain generalization. Cross-domain robustness and adaptability of 3D representations are crucial but not easily achieved through rotation augmentation. Motivated by the inherent advantages of intricate orientations in enhancing generalizability, we propose an innovative rotation-adaptive domain generalization framework for 3D point cloud analysis. Our approach aims to alleviate orientational shifts by leveraging intricate samples in an iterative learning process. Specifically, we identify the most challenging rotation for each point cloud and construct an intricate orientation set by optimizing intricate orientations. Subsequently, we employ an orientation-aware contrastive learning framework that incorporates an orientation consistency loss and a margin separation loss, enabling effective learning of categorically discriminative and generalizable features with rotation consistency. Extensive experiments and ablations conducted on 3D cross-domain benchmarks firmly establish the state-of-the-art performance of our proposed approach in the context of orientation-aware 3D domain generalization. © 2025, CC BY-NC-SA.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Anomaly detection mechanism based on hierarchical weights through large-scale log data

Anomaly detection mechanism based on hierarchical weights th...

引用

2021 International Conference on Computer Communication and Artificial Intelligence, CCAI 2021

作者： Xi, Lanlan Xin, Yang Luo, Shoushan Shang, Yanlei Tang, Qifeng University of Posts and Telecommunications School of Beijing Institute of Cyberspace Security Beijing China University of Postsand Telecommunications School of Beijing Institute of Cyberspace Security Beijing China Guizhou Provincial Key Laboratory of Public Big Data Guizhou University Beijing China National Engineering Laboratory for Big Data Distribution and Exchange Technologies Shanghai China China-Europe Digital Economy Expert Group Shanghai China

ISBN: (纸本)9781728194011

In order to realize intelligent Disaster Recovery and break the traditional reactive backup mode, it is necessary to forecast the potential system anomalies, and proactively backup the real-time datas and configurations. System logs record the running status as well as the critical events (including errors and warnings), which can help to detect system performance, debug system faults and analyze the causes of anomalies. What's more, with the features of real-time, hierarchies and easy-access, log data can be an ideal source for monitoring system status. To reduce the complexity and improve the robustness and practicability of existing log-based anomaly detection methods, we propose a new anomaly detection mechanism based on hierarchical weights, which can deal with unstable log data. We firstly extract semantic information of log strings, and get the word-level weights by SIF algorithm to embed log strings into vectors, which are then feed into attention-based Long Short-Term Memory(LSTM) deep learning network model. In addition to get sentence-level weight which can be used to explore the interdependence between different log sequences and improve the accuracy, we utilize attention weights to help with building workflow to diagnose the abnormal points in the execution of a specific task. Our experimental results show that the hierarchical weights mechanism can effectively improve accuracy of perdition task and reduce complexity of the model, which provides the feasibility foundation support for intelligent Disaster Recovery. © 2021 IEEE.

关键词： Anomaly detection

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

Towards a Smaller Student: Capacity Dynamic Distillation for...

引用

Conference on Computer Vision and Pattern Recognition (CVPR)

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

Previous Knowledge Distillation based efficient image retrieval methods employ a lightweight network as the stu-dent model for fast inference. However, the lightweight stu-dent 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 stu-dent model is gradually compressed during the training. To dynamically adjust the model capacity, our dynamic frame-work 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 accu-racy, 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. Code is avail-able at https://***/SCY-X/Capacity_Dynamic_Distillation.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Semi-supervised Deep Closest Point Method for Point Cloud Registration

Semi-supervised Deep Closest Point Method for Point Cloud Re...

引用

2021 International Conference on Robotics Automation and intelligent Control, ICRAIC 2021

作者： Wu, Yong Shen, Lingfeng Li, Ping Jiangsu Power Transmission and Transformation Co. Ltd. Nanjing211106 China College of Computer Science Nanjing University of Posts and Telecommunications Nanjing210023 China Jiangsu Key Laboratory of Big Data Security and Intelligent Processing Nanjing210023 China

Point cloud registration is one of the key issues in fields that need 3D scenes with global vision, including 3D scene reconstruction in robot technology, high-accuracy 3D map reconstruction in automatic driving, 3D reconstruction of real-time monitoring underground tunnel. Recently, point cloud registration methods based on deep neural networks have made great progress by learning from a large amount of well-aligned point cloud pairs. However, these large-scale labelled samples are usually difficult to obtain in the real-world applications such as visual simultaneous location and mapping of mobile robots which obtains unlabelled scene images in real time. To solve this problem, we propose a deep point cloud registration method based on the semi-supervised learning. Specifically, for the problem of less supervised data, we use Iterative Closest Point (ICP) algorithm to generate pseudo labels for the point cloud samples with unknown rigid transformations, effectively increasing the number of supervised data;in order to amend the pseudo labels, we design an alternating optimization algorithm to jointly learn the pseudo labels and deep point cloud registration model. The deep point cloud registration model and ICP are used alternately to continuously improve the precision of pseudo labels and the performance of the deep model. Experimental results show that this method can still obtain a reliable point cloud registration model even when just observing a small amount of point cloud samples with ground truth labels. © 2022 Institute of Physics Publishing. All rights reserved.

关键词： Surface measurement

来源：评论

学校读者我要写书评

暂无评论

Gnn-Mgrpool: Enhanced Graph Neural Networks with Multi-Granularity Pooling for Graph Classification

SSRN

引用

SSRN 2023年

作者： Sun, Haichao Wang, Guoyin Liu, Qun Guo, Yike Chongqing Key Laboratory of Computational Intelligence Chongqing University of Posts and Telecommunications Chongqing400065 China Key Laboratory of Big Data Intelligent Computing Chongqing University of Posts and Telecommunications Chongqing400065 China Key Laboratory of Cyberspace Big Data Intelligent Security Ministry of Education Chongqing University of Posts and Telecommunications Chongqing400065 China Department of Computer Science and Engineering The Hong Kong University of Science and Technology 999077 Hong Kong School of Computer Engineering and Science Shanghai University Shanghai200444 China

Graph neural networks (GNNs) have gained significant attention and have been applied in various domain tasks. Currently, numerous pooling approaches have been proposed to aggregate node features and obtain node embeddings. However, current GNNs are almost black-box models that typically use a flat or single pooling step to aggregate nodes by only considering the similarity between nodes within the cluster. These approaches ignore the influence of relationships within and between clusters. To address this issue, we propose a novel multi-granular pooling method to aggregate nodes by simultaneously considering density and relationships among nodes and clusters. This method allows to obtain multi-granular node embedding clusters. The clusters in the current layer are built upon the previous layer, and these clusters change from fine to coarse as the number of clusters decreases, which is achieved by using multi-granular pooling (MgrPool for short). Additionally, the node representation of each layer is established through the ratio of node distance within clusters to that of between clusters. Finally, we conduct several experiments on node and graph classification tasks by this pooling approach. The results demonstrate that our GNN-MgrPool model outperforms the state-of-the-art similar algorithms and largely improves the interpretability of the learning process. © 2023, The Authors. All rights reserved.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

Topic-aware Incentive Mechanism for Task Diffusion in Mobile Crowdsourcing through Social Network

引用

ACM Transactions on Internet Technology 2021年第1期22卷 1-23页

作者： Xu, Jia Zhou, Yuanhang Chen, Gongyu Ding, Yuqing Yang, Dejun Liu, Linfeng Jiangsu Key Laboratory of Big Data Security and Intelligent Processing Nanjing University of Posts and Telecommunications 9 Wenyuan Rd Jiangsu Nanjing210023 China Colorado School of Mines 1500 Illinois St. GoldenCO80401 United States

Crowdsourcing has become an efficient paradigm to utilize human intelligence to perform tasks that are challenging for machines. Many incentive mechanisms for crowdsourcing systems have been proposed. However, most of existing incentive mechanisms assume that there are sufficient participants to perform crowdsourcing tasks. In large-scale crowdsourcing scenarios, this assumption may be not applicable. To address this issue, we diffuse the crowdsourcing tasks in social network to increase the number of participants. To make the task diffusion more applicable to crowdsourcing system, we enhance the classic Independent Cascade model so the influence is strongly connected with both the types and topics of tasks. Based on the tailored task diffusion model, we formulate the Budget Feasible Task Diffusion (BFTD) problem for maximizing the value function of platform with constrained budget. We design a parameter estimation algorithm based on Expectation Maximization algorithm to estimate the parameters in proposed task diffusion model. Benefitting from the submodular property of the objective function, we apply the budget-feasible incentive mechanism, which satisfies desirable properties of computational efficiency, individual rationality, budget-feasible, truthfulness, and guaranteed approximation, to stimulate the task diffusers. The simulation results based on two real-world datasets show that our incentive mechanism can improve the number of active users and the task completion rate by 9.8% and 11%, on average. © 2021 Association for Computing Machinery.

关键词： Crowdsourcing

来源：评论

学校读者我要写书评

暂无评论

Indoor Navigation Mechanism based on 5G network for Large Parking Garage

Indoor Navigation Mechanism based on 5G network for Large Pa...

引用

International Conference on Computer Supported Cooperative Work in Design

作者： Xiaolong Xu Qun Ding Licheng Lin School of Computer Science Nanjing University of Posts and Telecommunications Nanjing China Jiangsu Key Laboratory Big Data Security and Intelligent Processing Nanjing University of Posts and Telecommunications Nanjing China

With the increase of car ownership in China, the problem of difficult parking in cities has become more and more serious. In large parking garage, finding an ideal parking space has become a daily problem for people; especially during the peak usage period of parking garage, a large number of vehicles will be driven into the parking garage, which will bring congestion, and at the same time, the demand of a large number of users in the area will lead to network congestion under 4G network, and users cannot access the network to check and find parking spaces. Therefore, this paper proposes a global optimal navigation mechanism with 5G network based on the above mentioned problems by analyzing the congestion situation during the peak usage of the parking garage. This mechanism uses an optimal parking space selection model and a global optimal scheduling model that can avoid congestion. Based on the model proposed in this paper, we construct a navigation planning system based on mobile, front-end and back-end. With the ultra-low latency, ultrahigh transmission efficiency and reliability of 5G network, the mobile port can select the optimal parking space according to the user's preference, and the back-end can show the parking space and congestion in the parking garage in real time according to the actual situation of the parking garage. Finally, the performance of the model and system proposed in this paper is verified through experiments.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Privacy Leak Detection Mechanism based on Service Binding

A Privacy Leak Detection Mechanism based on Service Binding

引用

International Symposium on System and Software Reliability (ISSSR)

作者： Boyang Wang Jinling He Yuanhan Du Ming Tang Xiaolong Xu Jiangsu Key Laboratory of Big Data Security & Intelligent Processing 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 mobile technology, users gradually store financial information and personal privacy data on mobile terminals, which makes mobile security particularly important. In order to effectively improve the detection speed of application privacy leaks, this paper proposes a concept of service binding, binding a minimum privacy permission set for each service provided by the application. Use the service binding concept to make the static analysis phase track only the sensitive data flow paths between APIs that apply the service's privacy permissions. In order to make up for the lack of static analysis to predict the behavior of application privacy leakage, this paper proposes a static analysis that effectively combines machine learning algorithms, abstracts the path characteristics of the static data streams obtained from static analysis into feature vectors, and further uses machine learning algorithms for learning classification. The experimental demonstration and performance analysis results show that the privacy leak detection mechanism based on service binding has faster detection speed and higher accuracy than FlowDroid.

关键词： Privacy data privacy Machine learning algorithms Accuracy Operating systems Static analysis Vectors

来源：评论

学校读者我要写书评

暂无评论

FP-WDDQN: An improved deep reinforcement learning algorithm for adaptive traffic signal control

FP-WDDQN: An improved deep reinforcement learning algorithm ...

引用

IEEE International Conference on data Mining Workshops (ICDM Workshops)

作者： Xiao Zhang Xiaolong Xu Jiangsu Key Laboratory of Big Data Security & Intelligent Processing Nanjing University of Posts and Telecommunications Nanjing China School of Computer Science Nanjing University of Posts and Telecommunications Nanjing China

Current adaptive traffic signal control methods based on centralized deep reinforcement learning are not applicable in large-scale adaptive traffic control environment. The scalability problem is overcome by assigning global control to each local RL agent through multi-intelligence reinforcement learning, but the environment now becomes partially visible ami non-stationarity from the perspective of each local agent due to limited communication between agents. In this paper, we propose a multi-agent framework called Forgetful Priority Weighed Double Deep Q-learning (FP-WDDQN). We firstly extend Weighted Double Deep Q-Learning(WDDQN) to the multi agent domain, so as to reduce the error caused by the algorithm's underestimation of the target network and get more accurate Q-value. Then we propose a new algorithm based on Forgetful Experience Mechanism(FEM) and Priority Experience Replay Mechanism(PERM). This mechanism makes WDDQN choose experience based on Temporal Difference (TD) error and Identify the importance of experience with FEM. It makes WDDQN preferentially select experience with high TD-error in the process of sampling and training its network based on FEM and TD-error, which improved the model training efficiency and stabilized the learning process. In this paper, we construct an urban traffic network with seven intersections in the simulation platform called Simulation of Urban MObility(SUMO), and the proposed FP-WDDQN is compared against Independent Double Deep Q-learning(IDDQN), WDDQN and Multi-agent Advantage Actor Critic (MA2C) under the condition of simulating the traffic dynamics of peak hours. Experiment shows FP-WDDQN has better performance than other algorithms in vehicle speed, intersection delay and intersection waiting queue length.

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：