检索结果-内蒙古大学图书馆

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Guo, Pengxin Zeng, Shuang Chen, Wenhao Zhang, Xiaodan Ren, Weihong Zhou, Yuyin Qu, Liangqiong School of Computing and Data Science The University of Hong Kong Hong Kong Department of Mathematics The University of Hong Kong Hong Kong College of Computer Science Beijing University of Technology China School of Mechanical Engineering and Automation Harbin Institute of Technology Shenzhen China Department of Computer Science and Engineering UC Santa Cruz United States

ISBN: (纸本)157735897X

Federated Learning (FL) aims to protect data privacy by enabling clients to collectively train machine learning models without sharing their raw data. However, recent studies demonstrate that information exchanged during FL is subject to Gradient Inversion Attacks (GIA) and, consequently, a variety of privacy-preserving methods have been integrated into FL to thwart such attacks, such as Secure Multi-party Computing (SMC), Homomorphic Encryption (HE), and Differential Privacy (DP). Despite their ability to protect data privacy, these approaches inherently involve substantial privacy-utility trade-offs. By revisiting the key to privacy exposure in FL under GIA, which lies in the frequent sharing of model gradients that contain private data, we take a new perspective by designing a novel privacy preserve FL framework that effectively "breaks the direct connection" between the shared parameters and the local private data to defend against GIA. Specifically, we propose a Hypernetwork Federated Learning (HyperFL) framework that utilizes hypernetworks to generate the parameters of the local model and only the hypernetwork parameters are uploaded to the server for aggregation. Theoretical analyses demonstrate the convergence rate of the proposed HyperFL, while extensive experimental results show the privacy-preserving capability and comparable performance of HyperFL. 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

Trillion Parameter AI Serving Infrastructure for Scientific Discovery: A Survey and Vision 23

Trillion Parameter AI Serving Infrastructure for Scientific ...

引用

Proceedings of the IEEE/ACM 10th International Conference on Big data Computing, Applications and Technologies

作者： Nathaniel C Hudson J. Gregory Pauloski Matt Baughman Alok Kamatar Mansi Sakarvadia Logan Ward Ryan Chard André Bauer Maksim Levental Wenyi Wang Will Engler Owen Price Skelly Ben Blaiszik Rick Stevens Kyle Chard Ian Foster Department of Computer Science University of Chicago Chicago Illinois United States Data Science and Learning Division Argonne National Laboratory Lemont Illinois US Data Science and Learning Division Argonne National Laboratory Lemont Illinois United States Globus University of Chicago Chicago Illinois United States Globus The University of Chicago Chicago Illinois United States Department of Computer Science University of Chicago Chicago Illinois USA

ISBN: (纸本)9798400704734

Deep learning methods are transforming research, enabling new techniques, and ultimately leading to new discoveries. As the demand for more capable AI models continues to grow, we are now entering an era of Trillion Parameter Models (TPM), or models with more than a trillion parameters---such as Huawei's PanGu-Σ. We describe a vision for the ecosystem of TPM users and providers that caters to the specific needs of the scientific community. We then outline the significant technical challenges and open problems in system design for serving TPMs to enable scientific research and discovery. Specifically, we describe the requirements of a comprehensive software stack and interfaces to support the diverse and flexible requirements of researchers.

关键词： artificial intelligence grid computing deep learning applications systems design survey

来源：评论

学校读者我要写书评

暂无评论

Interpretable Skill Learning for Dynamic Treatment Regimes through Imitation

Interpretable Skill Learning for Dynamic Treatment Regimes t...

引用

Annual Conference on Information sciences and Systems (CISS)

作者： Yushan Jiang Wenchao Yu Dongjin Song Wei Cheng Haifeng Chen Department of Computer Science and Engineering University of Connecticut Storrs CT USA NEC Laboratories American Data Science & System Security Princeton NJ USA

Imitation learning that mimics experts' skills from their demonstrations has shown great success in discovering dynamic treatment regimes, i.e., the optimal decision rules to treat an individual patient based on related evolving treatment and covariate history. Existing imitation learning methods, however, still lack the capability to interpret the underlying rationales of the learned policy in a faithful way. Moreover, since dynamic treatment regimes for patients often exhibit varying patterns, i.e., symptoms that transit from one to another, the flat policy learned by a vanilla imitation learning method is typically undesired. To this end, we propose an Interpretable Skill Learning (ISL) framework to resolve the aforementioned challenges for dynamic treatment regimes through imitation. The key idea is to model each segment of experts' demonstrations with a prototype layer and integrate it with the imitation learning layer to enhance the interpretation capability. On one hand, the ISL framework is able to provide interpretable explanations by matching the prototype to exemplar segments during the inference stage, which enables doctors to perform reasoning of the learned demonstrations based on human-understandable patient symptoms and lab results. On the other hand, the obtained skill embedding consisting of prototypes serves as conditional information to the imitation learning layer, which implicitly guides the policy network to provide a more accurate demonstration when the patients' state switches from one stage to another. Thoroughly empirical studies demonstrate that our proposed ISL technique can achieve better performance than state-of-the-art methods. Moreover, the proposed ISL framework also exhibits good interpretability which cannot be observed in existing methods.

关键词： Learning systems Visualization MIMICs Decision making Prototypes Medical services Cognition

来源：评论

学校读者我要写书评

暂无评论

Incentivized Federated Learning and Unlearning

引用

IEEE Transactions on Mobile Computing 2025年

作者： Ding, Ningning Sun, Zhenyu Wei, Ermin Berry, Randall Data Science and Analytics Thrust Internet of Things Thrust Information Hub Guangzhou511453 China Northwestern University Department of Electrical and Computer Engineering EvanstonIL60208 United States Northwestern University Electrical and Computer Engineering Department Industrial Engineering and Management Sciences Department EvanstonIL60208 United States

To protect users' right to be forgotten in federated learning, federated unlearning aims at eliminating the impact of leaving users' data on the global learned model. The current research in federated unlearning mainly concentrates on developing effective and efficient unlearning techniques. However, the issue of incentivizing valuable users to remain engaged and preventing their data from being unlearned is still under-explored, yet important to the unlearned model performance. This paper focuses on the incentive issue and develops an incentive mechanism for federated learning and unlearning. We first characterize the leaving users' impact on the global model accuracy and the required communication rounds for unlearning. Building on these results, we propose a four-stage game to capture the interaction and information updates during the learning and unlearning process. A key contribution is to summarize users' multi-dimensional private information into one-dimensional metrics to guide the incentive design. Interestingly, we prove that allowing federated unlearning can result in reduced payoffs for both the server and users, compared to a scenario without unlearning. Numerical results demonstrate the necessity of unlearning incentives for retaining valuable leaving users, and also show that our proposed mechanisms decrease the server's cost by up to 53.91% compared to state-of-the-art benchmarks. © 2002-2012 IEEE.

关键词： federated learning federated unlearning incentive mechanism

来源：评论

学校读者我要写书评

暂无评论

A churn prediction dataset from the telecom sector: a new benchmark for uplift modeling

arXiv

引用

arXiv 2023年

作者： Verhelst, Théo Mercier, Denis Shrestha, Jeevan Bontempi, Gianluca Machine Learning Group Department of Computer Science Université Libre de Bruxelles Brussels Belgium Data Science Team Orange Belgium Brussels Belgium

Uplift modeling, also known as individual treatment effect (ITE) estimation, is an important approach for data-driven decision making that aims to identify the causal impact of an intervention on individuals. This paper introduces a new benchmark dataset for uplift modeling focused on churn prediction, coming from a telecom company in Belgium, Orange Belgium. Churn, in this context, refers to customers terminating their subscription to the telecom service. This is the first publicly available dataset offering the possibility to evaluate the efficiency of uplift modeling on the churn prediction problem. Moreover, its unique characteristics make it more challenging than the few other public uplift datasets. © 2023, CC BY-NC-ND.

关键词： Decision making

来源：评论

学校读者我要写书评

暂无评论

Phishing Web Page Detection using Web Scraping

Phishing Web Page Detection using Web Scraping

引用

IEEE Southeastcon

作者： Mallika Boyapati Ramazan Aygun School of Data Science and Analytics Kennesaw State University Kennesaw GA USA Department of Computer Science Kennesaw State University Kennesaw GA USA

In today’s Internet era, webpages act as major user interfaces for the applications hosted by the organizations. One of the most common security attacks on the web page applications is phishing attacks. Phishing is a social engineering attack in which an adversary tries to steal user credentials by tricking them to believe that they are on a legitimate web page. Adversaries are using sophisticated and new ways to forge the web page designs craftily to trick the users into visiting the malicious links. The phishing webpages are used as a medium to carry out the art of phishing attacks. Web scraping is a methodology to extract features of each webpage. In this paper, web scraping is employed to extract hybrid feature set to implement Machine Learning (ML) models. The machine learning models like XGBoost, Multilayer Perceptron, Logistic Regression, SVM, Auto Encoder, Random Forest, Decision Trees, K-means, and Naive Bayes are built. All the models are tested with and without applying principal component analysis (PCA), a feature reduction technique. Extracting hybrid features by employing web scraping to train the ML models and finding the key features contributing to phishing detection on Phishpedia, Kaggle, and PhishTank datasets is the key contribution of this paper. Results show that XGBoost algorithm outperformed the all other classifiers with 98% accuracy or higher on the web scraped features on all the datasets. The model achieved higher precision and recall when compared to other approaches like CANTINA+, URL based approaches, SenseInput, Knowing Thy Domain, and Phishpedia on the Phishpedia dataset.

关键词： Uniform resource locators Visualization Machine learning algorithms Web page design Phishing Virtual assistants Web pages

来源：评论

学校读者我要写书评

暂无评论

Self-supervised Benchmark Lottery on ImageNet: Do Marginal Improvements Translate to Improvements on Similar datasets?

Self-supervised Benchmark Lottery on ImageNet: Do Marginal I...

引用

International Joint Conference on Neural Networks (IJCNN)

作者： Utku Ozbulak Esla Timothy Anzaku Solha Kang Wesley De Neve Joris Vankerschaver Center for Biosystems and Biotech Data Science Ghent University Global Campus Republic of Korea Department of Electronics and Information Systems Ghent University Belgium Department of Applied Mathematics Computer Science and Statistics Ghent University Belgium

ISBN: (数字)9798350359312

ISBN: (纸本)9798350359329

Machine learning (ML) research strongly relies on benchmarks in order to determine the relative effectiveness of newly proposed models. Recently, a number of prominent research effort argued that a number of models that improve the state-of-the-art by a small margin tend to do so by winning what they call a "benchmark lottery". An important benchmark in the field of machine learning and computer vision is the ImageNet where newly proposed models are often showcased based on their performance on this dataset. Given the large number of self-supervised learning (SSL) frameworks that has been proposed in the past couple of years each coming with marginal improvements on the ImageNet dataset, in this work, we evaluate whether those marginal improvements on ImageNet translate to improvements on similar datasets or not. To do so, we investigate twelve popular SSL frameworks on five ImageNet variants and discover that models that seem to perform well on ImageNet may experience significant performance declines on similar datasets. Specifically, state-of-the-art frameworks such as DINO and Swav, which are praised for their performance, exhibit substantial drops in performance while MoCo and Barlow Twins displays comparatively good results. As a result, we argue that otherwise good and desirable properties of models remain hidden when benchmarking is only performed on the ImageNet validation set, making us call for more adequate benchmarking. To avoid the "benchmark lottery" on ImageNet and to ensure a fair benchmarking process, we investigate the usage of a unified metric that takes into account the performance of models on other ImageNet variant datasets.

关键词： Measurement computer vision Computational modeling Neural networks Machine learning Self-supervised learning Benchmark testing

来源：评论

学校读者我要写书评

暂无评论

Parallelization of Software Systems Test Case Selection Algorithm Based on Singular Value Decomposition

arXiv

引用

arXiv 2022年

作者： Moghaddam, Mahdi Movahedian Department of Data and Computer Science Faculty of Mathematical Science Shahid Beheshti University Tehran Iran

When developing a software system, a change in one part of the system may lead to unwanted changes in other parts of the system. These affected parts may interfere with system performance, so regression testing is used to deal with these disorders. This test seeks to re-measure these sections to prevent these abnormalities, but it is difficult to identify these sections for re-examination. We try to cluster the changes of our software system based on the system functions by singular value decomposition, to be able to use to identify these parts during a new change, to perform the test again. In order to increase speedup, our calculations were performed in parallel on shared memory systems so that by increasing the scale of software systems, an optimal answer could be obtained. © 2022, CC BY.

关键词： Singular value decomposition

来源：评论

学校读者我要写书评

暂无评论

SVM-Enhanced Dynamic Audiology Screening with IoT-Driven Monitoring for Personalized Hearing Health Management 2

SVM-Enhanced Dynamic Audiology Screening with IoT-Driven Mon...

引用

2nd International Conference on Self Sustainable Artificial Intelligence Systems, ICSSAS 2024

作者： Ronald, Bindu S. Raman, Ramakrishnan Vekariya, Daxa Baskar, V. Vijaya Rahul, J. Srinivasan, C. Symbiosis Law School Maharashtra Pune India Maharashtra Pune India Parul Institute of Engineering and Technology Parul University Faculty of Engineering and Technology Gujarat Vadodara India Sathyabama Institute of Science and Technology School of Electrical and Electronics Engineering Tamil Nadu Chennai India Agni College of Technology Department of Artificial Intelligence and Data Science Tamil Nadu Chennai India Saveetha Institute of Medical and Technical Sciences Saveetha School of Engineering Department of Computer Science and Engineering Tamil Nadu Chennai India

ISBN: (纸本)9798350368413

The increasing prevalence of hearing impairment across diverse age demographics, particularly in elderly populations, highlights the need for early diagnosis and ongoing therapy. Traditional audiology screening techniques often lack the necessary dynamic flexibility to meet personalized hearing health requirements over time. This system introduces an innovative method for dynamic audiology screening via IoT-based monitoring, using Support Vector Machine (SVM) algorithms for individualized hearing health management. The incorporation of IoT devices allows ongoing, real-time auditory data acquisition from the user's surroundings, recording essential metrics such as sound frequencies, noise levels, and auditory reactions. This data is analyzed using SVM to categorize and identify probable hearing impairments, providing great precision in differentiating between various degrees of hearing loss. The SVM model's strong classification skills render it especially effective in differentiating between varying levels of hearing loss, facilitating early identification and tailored therapy strategies. This tailored method of audiology screening boosts diagnostic precision and facilitates proactive management, perhaps improving long-term results for individuals with hearing impairments. © 2024 IEEE.

关键词： Electronic health record

来源：评论

学校读者我要写书评

暂无评论

Evaluation of CAN Bus Security Vulnerabilities and Potential Solutions

Evaluation of CAN Bus Security Vulnerabilities and Potential...

引用

Women in data science at Prince Sultan University (WiDS PSU), International Conference of

作者： Asma Alfardus Danda B. Rawat Department of Electrical Engineering and Computer Science Data Science and Cybersecurity Center (DSC2) Howard University Washington D.C USA

The past decade has seen a significant increase in the automobile industry, which has come with some serious challenges and threats. Modem vehicles are now made up of complex mechanical systems, as well as sophisticated electronic devices and connections to the outside world. Various electronic devices utilize standard communication protocols, including the Controller Area Network (CAN), to establish communication with each other. Unfortunately, CAN lacks some fundamental security features, such as encryption and authentication, which makes it vulnerable to security attacks. This can lead to accidents and financial losses for the users of these vehicles. To address this issue, researchers have proposed a number of security measures, such as cryptography and Intrusion Detection Systems (IDS). This paper addresses the security vulnerabilities associated with CAN and proposes potential solutions to overcome its limitations.

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：