检索结果-内蒙古大学图书馆

2023 International Conference on networking and network Applications, NaNA 2023

作者： Zhang, Tao Yi Wang, YuBo Wei, ZhiCheng Key Lab of Network and Information Security China College of Computer and Cyberspace Security China Hebei Provincial Engineering Research Center for Supply Chain Big Data Analytics & Security Hebei Normal University Hebei Province Shijiazhuang China

ISBN: (纸本)9798350327380

Forecasting travel demand is a challenging task due to the complex spatial dependencies and dynamic temporal correlation of the traffic data. Furthermore, a limited representation of the given spatial graph structure may restrict the model's ability to effectively learn spatial-temporal dependencies. To extract latent semantic features from traffic data, this paper proposes a method to construct a traffic graph using the Markov cluster algorithm. The traffic semantic correlation matrix is constructed based on a traffic graph to obtain the deep semantic information. Specifically, this study proposes a Spatio-Temporal Graph Attention network(STGAT) for travel demand prediction. STGAT uses a graph attention layer based on the Node2Vec graph embedding algorithm, two convolutional layers based on the Markov cluster algorithm, and a long short-term memory network to capture spatial-temporal dependencies. Experimental results on the NYC Taxi dataset and the Chengdu online car dataset demonstrate that STGAT achieved state-of-the-art performance compared to other baseline models. © 2023 IEEE.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

Advancing MobileNet security: Weighted Adversarial Learning in Convolutional Neural networks 11

Advancing MobileNet Security: Weighted Adversarial Learning ...

引用

11th International Conference on Machine Intelligence Theory and Applications, MiTA 2024

作者： Quadri, Hakeem Gu, Bruce Wang, Hua Islam, Sardar Victoria University Department of Information Technology Melbourne Australia Qilu University of Technology Key Laboratory of Computing Power Network and Information Security Jinan China Victoria University Institute for sustainable Industries and Liveable Cities Melbourne Australia

ISBN: (纸本)9798350385670

Convolutional Neural networks (CNN), particularly low latency models such as MobileNets have excelled in many applications, including object detection in images, speech recognition and natural language processing, but they are vulnerable to subtle perturbations that are virtually imperceptible to the human eye yet can deceive the network into misclassifying images. To enhance the robustness of such CNNs against adversaries, conventional adversarial training methods treat all data points as equally important and susceptible to attack. A weighted adversarial learning algorithm is developed in a Stackelberg game framework. This approach prioritizes data points that are more susceptible to attacks during network training. To further optimize the algorithm, we employ a Reinforcement Learning (R L) agent to fine-tune the hyperparameters of the model, thereby increasing its robustness. Our findings indicate an increased robustness of 66.18 % of the Weighted Adversary Reinforced Stackelberg Learning (WARS) against the traditional adversarial training of 64.72 % in a one epoch training, using the CIFAR-10 dataset. We conclude that the WARS represents a valuable adversarial training method for bolstering the robustness of low latency CNN models. © 2024 IEEE.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

Mal-lightDet: A light method to detect malicious encrypted traffic based on machine learning 23

Mal-lightDet: A light method to detect malicious encrypted t...

引用

2023 4th International Conference on Control, Robotics and Intelligent System, CCRIS 2023

作者： Liu, Xu Liu, Zhang Zhao, Xiaojing Ren, Zhengwei School of Computer Science and Technology Changchun University of Science and Technology China Jilin Province Key Laboratory of Network and Information Security China School of Artificial Intelligence Changchun University of Science and Technology Changchun China

ISBN: (纸本)9798400708190

Encryption not only protects the network security and privacy, but also encrypts attackers' malicious traffic to evade detection. Thus, how to detect malicious encrypted traffic is critical for network security. From the perspective of privacy, methods based on ciphertext seem to be more attractive than decryption-based methods. However, due to performance limitation or privacy violations, most existing methods are far from satisfactory. Machine learning models provide a solution to the malicious encrypted traffic detection, but they place great demand on data features. To balance the detection performance and feature effectiveness, this paper proposes a framework called Mal-lightDet for light detection of malicious encrypted traffic based on machine learning. Mal-lightDet includes a feature-group selection model that mines the relevance between features to represent the traffic flow, which is less considered in the existing methods, and a voting-based detection model to eliminate the bias of different machine learning models. The selected feature subset not only preserves the detection effectiveness but also reduces the computation brought by massive training data. The experimental results prove the efficiency and stability of Mal-lightDet to distinguish malicious flow data by comparing it with other state-of-art methods. Besides, Mal-lightDet performs faster, more accurately, and more stably than some deep learning methods. © 2023 ACM.

关键词： network security

来源：评论

学校读者我要写书评

暂无评论

Ultra-high definition video quality assessment method based on deep learning 9

Ultra-high definition video quality assessment method based ...

引用

9th IEEE International Conference on Data Science in Cyberspace, DSC 2024

作者： Li, Wei He, Can Zhou, MingLe Zhang, HengYu Liu, WenLong Hisense Video Technology Co. Ltd. Qing Dao China Qilu University of Technology Shandong Academy of Sciences Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center National Supercomputer Center in Jinan Jinan China Shandong Fundamental Research Center for Computer Science Shandong Provincial Key Laboratory of Computer Networks Jinan China

ISBN: (纸本)9798350391367

With the development of display technology, people's demand for high-definition video is increasing, and the requirements for video quality are also increasing. Nowadays, a large number of videos have emerged, and how to distinguish the quality of videos faces many challenges. Nowadays, deep learning is developing rapidly, and many video quality assessment (VQA) methods have emerged. These VQA methods can efficiently evaluate video quality and save a lot of manpower and material resources. To achieve the evaluation of UHD video quality, we propose an inter-frame downsampling UHD video quality assessment model (IFDU-VQA) based on deep learning. This model is a no-reference (NR) evaluation model that adopts a dual-branch structure: one branch uses a recursive network to deeply extract video inter-frame features, and combines inter-frame downsampling technology with a multi-information fusion (MIF) module to achieve different time and frequency information. Effective fusion;the other branch applies convolutional neural networks (CNN) to extract features from individual video frames. By fusing the outputs of the two branches, regression and pooling modules are used to calculate the final quality score of the video. Compared with multiple existing models, experimental results show that IFDU-VQA demonstrates superior performance on multiple standard data sets. © 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

Discourse Relation-Aware Multi-turn Dialogue Response Generation 12th

Discourse Relation-Aware Multi-turn Dialogue Response Genera...

引用

12th National CCF Conference on Natural Language Processing and Chinese Computing, NLPCC 2023

作者： Wang, Huijie He, Ruifang Jia, Yungang Xu, Jing Wang, Bo College of Intelligence and Computing Tianjin University Tianjin China Tianjin Key Laboratory of Cognitive Computing and Application Tianjin China Tianjin Branch of National Computer Network and Information Security Management Center Tianjin China

ISBN: (纸本)9783031446924

Multi-turn dialogue response generation aims to generate a response with consideration of the context. It is not equal to multiple single-turn dialogues due to the context dependence of response. Many existing models achieve great success for response generation, but they still struggle to model the contextual semantics of dialogue history. Sequence models have difficulties to explore the interactive relations between contextual utterances, which affects the coherence of generated responses. To solve the issue, we propose a discourse relation-aware model, which encodes the contextual utterances with a directed acyclic graph neural network (DAGNN) with constraints on dialogue-specific discourse relations to better model the intrinsic structure. Besides, we introduce an auxiliary discourse relation recognition task to enhance the model’s ability of representing the context. Extensive experimental results show that our proposed model outperforms baselines. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Directed graphs

来源：评论

学校读者我要写书评

暂无评论

Convertible certificateless proxy re-signcryption for data access control in cloud computing

Convertible certificateless proxy re-signcryption for data a...

引用

2023 International Conference on computer Application and information security, ICCAIS 2023

作者： Wu, Chenhuang Yang, Linghui Yuan, Beibei Li, Huimin Huang, Hui Fujian Key Laboratory of Financial Information Processing Putian University Fujian Putian351100 China Fujian Zhongxin Network Security Information Technology Co. Ltd. Fujian Fuzhou350011 China School of Computer Science Minnan Normal University Fujian Zhangzhou363000 China

ISBN: (纸本)9781510675032

Cloud storage service enables users not only to save their local data storage costs but also to access their outsourced data at any time and anywhere. Traditionally, a data owner outsources the encrypted data to the cloud server and then checks its integrity with the help of cloud auditing strategies. Obviously, in such a manner, only confidentiality and integrity are considered following the above two independent steps. Furthermore, it becomes an issue that the data owner may want to share their outsourced data with some authorized data receivers and convince them that the data are from and the same as the data owner's. To solve this issue, absorbing the advantage of the certificateless cryptosystem, we propose an efficient certificateless proxy re-signcryption (CLPRSC) for data access control in cloud computing. Moreover, the CLPRSC possesses ciphertext verifiable, convertible, and designated re-signcryptable properties with the highest efficiency among the achievable signcryption with proxy re-encryption schemes, which may be of independent interest. © 2024 SPIE.

关键词： Access control

来源：评论

学校读者我要写书评

暂无评论

Open-vocabulary 3D Semantic Understanding via Affinity Neural Radiance Fields 6

Open-vocabulary 3D Semantic Understanding via Affinity Neura...

引用

6th International Conference on Next Generation Data-Driven networks, NGDN 2024

作者： Fan, Yujie Luo, Huan Computer and Data Science Fuzhou University Fuzhou China Fujian Provincial Key Laboratory of Network Computing and Intelligent Information Processing Fuzhou University Fuzhou China

ISBN: (纸本)9798350388374

Three-dimensional semantic understanding using only several multi-view images can largely reduce the communication burden on the network. In addition, while point clouds are extensively studied for 3D scene understanding, utilization of multi-view image data offers rich visual details and texture information. However, challenges persist in lifting 2D semantic features to 3D space and leveraging language for segmentation. Inspired by recent advancements, this paper proposes a method that combines CLIP features and SAM masks to create a feature field capable of segmenting objects via natural language text across 2D multi-view and 3D Gaussian splatting. It offers a promising function for extracting 3D assets for game engines and the metaverse. Our method involves mask generation from video frames, extracting physical scales via RGB Nerf with masks, and organizing hierarchical information for semantic comprehension. In the training process, affinity features maintain scale properties and guide CLIP feature generation with auto weights blending for semantic robustness. A straightforward 3D splatting CLIP feature approach and canonical text methodology enhance query robustness across 2D multi-view and 3D splatting through relevance score calculation based on text CLIP features for inference. Experimental results demonstrate promising improvements in semantic understanding of 3D scenes. © 2024 IEEE.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Multi-instance partial-label learning: towards exploiting dual inexact supervision

引用

Science China(information Sciences) 2024年第3期67卷 48-61页

作者： Wei TANG Weijia ZHANG Min-Ling ZHANG School of Computer Science and Engineering Southeast University Key Laboratory of Computer Network and Information Integration (Southeast University) Ministry of Education School of Information and Physical Sciences The University of Newcastle

Weakly supervised machine learning algorithms are able to learn from ambiguous samples or labels, e.g., multi-instance learning or partial-label learning. However, in some real-world tasks, each training sample is associated with not only multiple instances but also a candidate label set that contains one groundtruth label and some false positive labels. Specifically, at least one instance pertains to the ground-truth label while no instance belongs to the false positive labels. In this paper, we formalize such problems as multi-instance partial-label learning(MIPL). Existing multi-instance learning algorithms and partial-label learning algorithms are suboptimal for solving MIPL problems since the former fails to disambiguate a candidate label set, and the latter cannot handle a multi-instance bag. To address these issues, a tailored algorithm named MIPLGP, i.e., multi-instance partial-label learning with Gaussian processes, is *** first assigns each instance with a candidate label set in an augmented label space, then transforms the candidate label set into a logarithmic space to yield the disambiguated and continuous labels via an exclusive disambiguation strategy, and last induces a model based on the Gaussian processes. Experimental results on various datasets validate that MIPLGP is superior to well-established multi-instance learning and partial-label learning algorithms for solving MIPL problems.

关键词： machine learning multi-instance partial-label learning multi-instance learning partial-label learning Gaussian processes

来源：评论

学校读者我要写书评

暂无评论

TMOVF: A Task-Agnostic Model Ownership Verification Framework

TMOVF: A Task-Agnostic Model Ownership Verification Framewor...

引用

2023 IEEE International Conference on Systems, Man, and Cybernetics, SMC 2023

作者： Sun, Zhe Huang, Zhongyu Zhao, Wangqi Zhang, Yu Zhang, Jianzhong College of Cyber Science College of Computer Science Nankai University Tianjin China Tianjin Key Laboratory of Network and Data Security Technology Tianjin China

ISBN: (纸本)9798350337020

The protection of model intellectual property is becoming an increasingly important issue. However, the existing methods for protecting model ownership, although effective, have limitations. Firstly, they primarily focus on classification models, and secondly, most of the proposed methods reduce the model's utility. To overcome these shortcomings, this paper proposes a task-agnostic model ownership verification framework based on feature fingerprint, called TMOVF, which separates ownership verification from model task. Our key idea is that model knowledge can be uniquely characterized by the extracted features, which may be high-dimensional, complicated, and difficult to compare for each input sample. Nevertheless, these features contain inherent information that cannot be ignored in cases of piracy. To measure the inheritance of our fingerprint, we introduce outlier detection into model ownership verification, which is a first in the field. By reconstructing the outlier detection algorithm, we extract the feature fingerprints of the victim model and the suspicious model, and compute the outliers of their feature fingerprints. By comparing the results, we can verify the ownership of the models. We conduct extensive experiments to evaluate our framework and demonstrate the inheritability of feature fingerprints in stolen models. Our experiments show that the framework is effective in verifying ownership, regardless of the model task. Additionally, our results demonstrate that our framework is more effective than existing methods. © 2023 IEEE.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Text information Mining in Cyberspace: An information Extraction Method Based on T5 and keyBERT 9

Text Information Mining in Cyberspace: An Information Extrac...

引用

9th IEEE International Conference on Data Science in Cyberspace, DSC 2024

作者： Liu, Kuan Li, Yumei Qi, Yuanhua Qi, Ning Zhai, Mengran Qilu University of Technology Shandong Academy of Sciences Information Research Institute Shandong Academy of Sciences Jinan China Qilu University of Technology Shandong Academy of Sciences Shandong Institute of Economy and Informatization Development Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center National Supercomputer Center in Jinan Jinan China Qilu University of Technology Shandong Academy of Sciences Shandong Provincial Key Laboratory of Computer Networks Shandong Fundamental Research Center for Computer Science Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center National Supercomputer Center in Jinan Jinan China SHANDONG SCICOM Information and Economy Research Institute Co. Ltd Jinan China Shandong Institute of Economy and Informatization Development Jinan China

ISBN: (纸本)9798350391367

With the development of the digital economy and the advent of the big data era, the rapid growth of textual information in cyberspace has placed higher demands on information processing technology. This paper proposes an information extraction technique based on the Text-to-Text Transfer Transformer, known as T5, and keyBERT models, aiming to efficiently distill the key content from textual information in cyberspace. This method combines automatic summarization and keyword extraction to form information briefs. The experiments selected representative policy documents related to data elements as input data and employed information entropy and ROUGE scores as metrics to evaluate the automatic summarization model. The results were compared with the outputs of similar models. Experimental results indicate that the T5 model outperforms other models in terms of summarization effectiveness, and the proposed information extraction method shows significant advantages in readability and processing efficiency, demonstrating practical application value. © 2024 IEEE.

关键词： Spatio-temporal data

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：