检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Sun, Hong Li, Ziqiang Liu, Lei Li, Bin University of Science and Technology of China Hefei230027 China Laboratory for Big Data and Decision Zhejiang Lab. Zhejiang Hangzhou311121 China

The proliferation of malicious deepfake applications has ignited substantial public apprehension, casting a shadow of doubt upon the integrity of digital media. Despite the development of proficient deepfake detection mechanisms, they persistently demonstrate pronounced vulnerability to an array of attacks. It is noteworthy that the pre-existing repertoire of attacks predominantly comprises adversarial example attack, predominantly manifesting during the testing phase. In the present study, we introduce a pioneering paradigm denominated as"Bad-Deepfake," which represents a novel foray into the realm of backdoor attacks levied against deepfake detectors. Our approach hinges upon the strategic manipulation of a delimited subset of the training data, enabling us to wield disproportionate influence over the operational characteristics of a trained model. This manipulation leverages inherent frailties inherent to deepfake detectors, affording us the capacity to engineer triggers and judiciously select the most efficacious samples for the construction of the poisoned set. Through the synergistic amalgamation of these sophisticated techniques, we achieve an remarkable performance—a 100% attack success rate (ASR) against extensively employed deepfake detectors. © 2024, CC BY.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

CREAM: Named Entity Recognition with Concise query and REgion-Aware Minimization 24th

CREAM: Named Entity Recognition with Concise query and REg...

引用

24th International Conference on Web Information Systems Engineering, WISE 2023

作者： Yao, Xun Yang, Qihang Hu, Xinrong Yang, Jie Guo, Yi School of Computer Science and Artificial Intelligence Wuhan Textile University Wuhan China School of Computing and Information Technology University of Wollongong Wollongong Australia School of Computer Data and Mathematical Sciences Western Sydney University Sydney Australia

ISBN: (纸本)9789819972531

Recent advancements in Machine Reading Comprehension (MRC) models have sparked interest in the field of Named Entity Recognition (NER), where entities are extracted as answers of given queries. Yet, existing MRC-based models face several challenges, including high computational costs, limited consideration of entity content information, and the tendency to generate sharp boundaries, that hinder their generalizability. To alleviate these issues, this paper introduces CREAM, an enhanced model leveraging Concise query and REgion-Aware Minimization. First, we propose a simple yet effective strategy of generating concise queries based primarily on entity categories. Second, we propose to go beyond existing methods by identifying entire entities, instead of just their boundaries (start and end positions), with an efficient continuous cross-entropy loss. An in-depth analysis is further provided to reveal their benefit. The proposed method is evaluated on six well-known NER benchmarks. Experimental results demonstrate its remarkable effectiveness by surpassing the current state-of-the-art models, with the substantial averaged improvement of 2.74, 1.12, and 2.38 absolute percentage points in Precision, Recall, and F1 metrics, respectively. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Machine Reading Comprehension Named Entity Recognition Query Optimization Region-Aware Loss

来源：评论

学校读者我要写书评

暂无评论

Dynamic Adapter with Semantics Disentangling for Cross-lingual Cross-modal Retrieval

arXiv

引用

arXiv 2024年

作者： Cai, Rui Dong, Zhiyu Dong, Jianfeng Wang, Xun The College of Computer Science and Technology Zhejiang Gongshang University Hangzhou China Zhejiang Key Laboratory of Big Data and Future E-Commerce Technology Hangzhou China

Existing cross-modal retrieval methods typically rely on large-scale vision-language pair data. This makes it challenging to efficiently develop a cross-modal retrieval model for under-resourced languages of interest. Therefore, Cross-lingual Cross-modal Retrieval (CCR), which aims to align vision and the low-resource language (the target language) without using any human-labeled target-language data, has gained increasing attention. As a general parameter-efficient way, a common solution is to utilize adapter modules to transfer the vision-language alignment ability of Vision-Language Pretraining (VLP) models from a source language to a target language. However, these adapters are usually static once learned, making it difficult to adapt to target-language captions with varied expressions. To alleviate it, we propose Dynamic Adapter with Semantics Disentangling (DASD), whose parameters are dynamically generated conditioned on the characteristics of the input captions. Considering that the semantics and expression styles of the input caption largely influence how to encode it, we propose a semantic disentangling module to extract the semantic-related and semantic-agnostic features from the input, ensuring that generated adapters are well-suited to the characteristics of input caption. Extensive experiments on two image-text datasets and one video-text dataset demonstrate the effectiveness of our model for cross-lingual cross-modal retrieval, as well as its good compatibility with various VLP models. Code - https://***/HuiGuanLab/DASD. © 2024, CC BY.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

A Multiple Indicator Credit Ranking Scheme Based on Notary Mechanism

A Multiple Indicator Credit Ranking Scheme Based on Notary M...

引用

IEEE International Symposium on Dependable, Autonomic and Secure Computing (DASC)

作者： Lang Chen Yuling Chen Chaoyue Tan Hui Dou Yun Luo Pengyu Chen State Key Laboratory of Public Big Data College of Computer Science and Technology Guizhou University Guiyang China

Blockchain has penetrated multiple fields such as finance, healthcare, supply chain, and intelligent transportation. However, the value exchange between different independent blockchain systems limits the expansion of blockchain. Cross-chain technology can achieve the transfer of digital assets between blockchain networks, and the notary mechanism is a typical cross-chain technology that can achieve value exchange between different blockchains. However, it requires trust in specific notaries. To address these issues, this paper proposes a multi-indicator credit ranking scheme based on notaries to improve the reliability of notaries and ensure the security of value exchange between different blockchains. To tackle the problem of single and incomplete inherent value evaluation indicators in traditional notary mechanisms, a multi-indicator evaluation is adopted to ensure the fairness of notary node credit evaluation. To address the issue of inaccurate comprehensive credit evaluation of notaries, a comprehensive evaluation and ranking of notary nodes are conducted based on their historical transaction information to improve the accuracy of their credit evaluation. The experimental results show that using multiple indicators to evaluate the inherent value of notary nodes can comprehensively evaluate the scheme and accurately screen out high-credit nodes to act as notaries, improving the security of cross-chain asset transactions.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Subset Selection in Support Vector Machine

Subset Selection in Support Vector Machine

引用

International Joint Conference on Neural Networks (IJCNN)

作者： Jiangshuo Zhao Canhong Wen School of Data Science University of Science and Technology of China Hefei China Department of Statistics and Finance School of Management University of Science and Technology of China Hefei China

ISBN: (数字)9798350359312

ISBN: (纸本)9798350359329

Support Vector Machines (SVMs) face a significant challenge when dealing with high-dimensional datasets, as their performance can be severely compromised by the inclusion of numerous redundant variables. To address this challenge, we propose a novel approach that integrates ℓ 0 and ℓ 2 penalties to enhance the performance of SVM in high-dimensional settings. The ℓ 0 penalty is introduced for best subset selection, addressing the issue of redundancy. In addition, we incorporate an ℓ 2 penalty in our proposal, which effectively reduces the impact of noise, thereby improving the stability and generalization capability of the model. To solve the underlying optimization problem, we develop an efficient algorithm that utilizes primaldual hybrid gradient method and the concept of splicing to reach a stable solution. Through simulation studies, we demonstrate the superior performance of our proposal compared to the state-of-the-art methods in terms of prediction accuracy and variables selection. To further validate the effectiveness of our method, we conduct experiments using three classification datasets from the NIPS 2003 Feature Selection Challenge. The results show the benefits of our proposed method in selecting variables and achieving more accurate predictions.

关键词： Support vector machines Gradient methods Accuracy Splicing Redundancy Prediction algorithms Vectors

来源：评论

学校读者我要写书评

暂无评论

Facial Recognition for Humanitarian Efforts: A Deep Learning based Solution for Missing Person Identification

Facial Recognition for Humanitarian Efforts: A Deep Learning...

引用

Advances in Computing, Communication and Applied Informatics (ACCAI), International Conference on

作者： Dhruv Limbani Abhishek Barhate D Hemavathi Department of Data Science and Business Systems School of Computing SRM Institute of Science and Technology Kattankulathur Chennai India

ISBN: (数字)9798350389449

ISBN: (纸本)9798350389456

Our work offers a novel method for identifying missing persons that makes use of recent developments in face recognition technology. Conventional techniques depend on converting images into embeddings and then figuring out how far apart picture embeddings are from one another by calculating distances between them, making it computationally demanding, particularly while dealing with huge databases. Our process involves first extracting faces from images using OpenCV’s face extractor module, and then using VGGFace to turn those faces into embeddings. However, we employ Principal Component Analysis (PCA) to further reduce their dimensionality to 128-dimensional vectors rather than directly comparing embeddings. On the basis of these reduced embeddings, a Dense Neural Network is then trained. With this method, the traditional O(n) time for classification is greatly reduced to O(1). Our tests show that our methodology outperforms numerous other methods and reaches an excellent accuracy of 98.75%. This simplified method gives encouraging outcomes in situations involving missing persons identification in addition to increasing efficiency.

关键词： Deep learning Adaptation models Accuracy Face recognition Computational modeling Predictive models Vectors

来源：评论

学校读者我要写书评

暂无评论

Sustainability Forecasting for Deep Learning Packages 31

Sustainability Forecasting for Deep Learning Packages

引用

31st IEEE International Conference on Software Analysis, Evolution and Reengineering, SANER 2024

作者： Han, Junxiao Wang, Yunkun Liu, Zhongxin Bao, Lingfeng Liu, Jiakun Lo, David Deng, Shuiguang School of Computer & Computing Science Hangzhou City University Hangzhou China College of Computer Science and Technology Zhejiang University Hangzhou China Zhejiang University The State Key Laboratory of Blockchain and Data Security Hangzhou China School of Computing and Information Systems Singapore Management University Singapore Singapore

ISBN: (纸本)9798350330663

Deep Learning (DL) technologies have been widely adopted to tackle various tasks. In this process, through software dependencies, a multi-layer DL supply chain (SC) is formed, with DL frameworks acting as the root, DL packages acting as the bridge nodes, and downstream DL projects acting as the periphery. However, most Open Source Software (OSS) projects may fail. Considering the crucial position of DL packages in the DL SC, to foster the sustainable development of DL SCs and DL packages, we aim to forecast the long-term sustainability of DL packages. Here, sustained activity is adopted as the main proxy of sustainability, and the sustainability status is classified as 'sus-tainable' or 'dormant'. Relatedly, a DL package is considered as 'sustainable' if it has sustained activity in its last 12 months. Otherwise, it is deemed as 'dormant'. To this end, we propose an approach that begins with obtaining longitudinal features for each DL package in each month. Then, we develop a model to forecast the sustainability of DL packages by incorporating the longitudinal features, which can aptly predict sustainability with an accuracy of up to 0.81. Subsequently, an interpretable module is developed to interpret the determinants (i.e., important features) that impact the sustainability of DL packages. Finally, we generate sustainability trajectories for each DL package to better understand the monthly changes of their sustainability status. Our findings uncover that for most DL packages, fewer but more centralized developers and a balanced collaboration are more likely to help sustain the DL packages. Furthermore, although some DL packages are sustainable, their sustainability trajectories present statistically decreasing trends over time. Based on the findings, we shed light on the dynamic sustainability of DL packages, highlight future research directions, and provide practical suggestions to DL package maintainers, developers, users, and software engineering researchers. ©

关键词： Open source software

来源：评论

学校读者我要写书评

暂无评论

LEVERAGE CLASSIFIER: ANOTHER LOOK AT SUPPORT VECTOR MACHINE

arXiv

引用

arXiv 2023年

作者： Han, Yixin Yu, Jun Zhang, Nan Meng, Cheng Ma, Ping Zhong, Wenxuan Zou, Changliang School of Statistics and Data Science LPMC & KLMDASR Nankai University Tianjin China School of Mathematics and Statistics Beijing Institute of Technology Beijing China School of Data Science Fudan University Shanghai China Institute of Statistics and Big Data Renmin University Beijing China Department of Statistics University of Georgia AthensGA United States

Support vector machine (SVM) is a popular classifier known for accuracy, flexibility, and robustness. However, its intensive computation has hindered its application to large-scale datasets. In this paper, we propose a new optimal leverage classifier based on linear SVM under a nonseparable setting. Our classifier aims to select an informative subset of the training sample to reduce data size, enabling efficient computation while maintaining high accuracy. We take a novel view of SVM under the general subsampling framework and rigorously investigate the statistical properties. We propose a two-step subsampling procedure consisting of a pilot estimation of the optimal subsampling probabilities and a subsampling step to construct the classifier. We develop a new Bahadur representation of the SVM coefficients and derive unconditional asymptotic distribution and optimal subsampling probabilities without giving the full sample. Numerical results demonstrate that our classifiers outperform the existing methods in terms of estimation, computation, and prediction. Copyright © 2023, The Authors. All rights reserved.

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Delayed Antarctic melt season reduces albedo feedback

引用

National science Review 2023年第9期10卷 236-246页

作者： Lei Liang Huadong Guo Shuang Liang Xichen Li John C.Moore Xinwu Li Xiao Cheng Wenjin Wu Yan Liu Annette Rinke Gensuo Jia Feifei Pan Chen Gong Key Laboratory of Digital Earth Science Aerospace Information Research Institute Chinese Academy of Sciences International Research Center of Big Data for Sustainable Development Goals Institute of Atmospheric Physics Chinese Academy of Sciences College of Global Change and Earth System Science Beijing Normal University Arctic Centre University of Lapland CAS Center for Excellence in Tibetan Plateau Earth Sciences School of Geospatial Engineering and Science Sun Yat-sen University Alfred Wegener Institute Helmholtz Centre for Polar and Marine Research Department of Geography University of North Texas

Antarctica' s response to climate change varies greatly both spatially and *** melting impacts mass balance and also lowers surface *** use a 43-year record(from 1978 to 2020) of Antarctic snow melt seasons from space-borne microwave radiometers with a machine-learning algorithm to show that both the onset and the end of the melt season are being ***-causality analysis shows that melt end is delayed due to increased heat flux from the ocean to the atmosphere at minimum sea-ice extent from warming *** onset is Granger-caused primarily by the turbulent heat flux from ocean to atmosphere that is in turn driven by sea-ice *** snowmelt season leads to a net decrease in the absorption of solar irradiance,as a delayed summer means that higher albedo occurs after the period of maximum solar radiation,which changes Antarctica's radiation balance more than sea-ice cover.

关键词： Antarctic ice sheet snowmelt season delay sea ice solar irradiance

来源：评论

学校读者我要写书评

暂无评论

Comments on "A Lightweight Privacy Preserving Authentication Protocol for VANETs" 9

Comments on "A Lightweight Privacy Preserving Authentication...

引用

9th International Conference on Computer and Communications, ICCC 2023

作者： Awais, Syed Muhammad Yucheng, Wu Mahmood, Khalid Kharel, Rupak Chongqing University School of Microelectronics and Communication Engineering Chongqing China University of Central Lancashire School of Psychology and Computer Science Graduate School of Intelligent Data Science Preston United Kingdom National Yunlin University of Science and Technology School of Psychology and Computer Science Graduate School of Intelligent Data Science Yunlin64002 Taiwan University of Central Lancashire School of Psychology and Computer Science Faculty of Electrical and Electronics Engineering Preston United Kingdom Ton Duc Thang University School of Psychology and Computer Science Faculty of Electrical and Electronics Engineering Ho Chi Minh City Viet Nam

ISBN: (纸本)9798350317251

Vehicle use and the concept of a "smart city"are developing quickly. As a result of this progression, the Vehicular Ad-Hoc Network (VANET) is a popular network for inter-vehicular communication. The data gathering about the state of the road, the whereabouts of vehicles, their speeds, and the description of traffic congestion are obtained using the VANET. The public nature of information flow on the VANET creates serious security risks. data security is typically one of the most important duties of VANET. As a result, it becomes one of the researchers' top priorities. Li et al. [1] introduced a Lightweight Privacy Preserving Authentication Protocol for VANETs. They stated that several attacks could be resisted using their secure authentication devised protocol. However, after analyzing we discovered that their authentication protocol is susceptible to roadside impersonation attack and it does not provide vehicle anonymity. In the end, we have provided a few suggestions to cope with Li et al.'s devised protocol current shortcomings. © 2023 IEEE.

关键词： Vehicular ad hoc 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：