Pub Date : 2024-01-06DOI: 10.1109/CCNC51664.2024.10454840
Francesco Colosimo, F. Rango
New machine learning (ML) paradigms are being researched thanks to the current widespread adoption of AI-based services. Since it enables several users to cooperatively train a global model without disclosing their local training data, Federated Learning (FL) represents a new distributed methodology capable of attaining stronger privacy and security guarantees than current methodologies. In this paper, a study of the properties of FL is conducted, with an emphasis on security issues. In detail, a thorough investigation of currently known vulnerabilities and their corresponding countermeasures is conducted, focusing on aggregation algorithms that provide robustness against Byzantine failures. Following this direction, new aggregation algorithms are observed on a set of simulations that recreate realistic scenarios, in the absence and presence of Byzantine adversaries. These combine the Distance-based Krum approach with the Statistical based aggregation algorithm. Achieved results demonstrate the functionality of the proposed solutions in terms of accuracy and convergence rounds in comparison with well-known federated algorithms under a correct and incorrect estimation of the attackers number.
{"title":"Distance-Statistical Based Byzantine-Robust Algorithms in Federated Learning","authors":"Francesco Colosimo, F. Rango","doi":"10.1109/CCNC51664.2024.10454840","DOIUrl":"https://doi.org/10.1109/CCNC51664.2024.10454840","url":null,"abstract":"New machine learning (ML) paradigms are being researched thanks to the current widespread adoption of AI-based services. Since it enables several users to cooperatively train a global model without disclosing their local training data, Federated Learning (FL) represents a new distributed methodology capable of attaining stronger privacy and security guarantees than current methodologies. In this paper, a study of the properties of FL is conducted, with an emphasis on security issues. In detail, a thorough investigation of currently known vulnerabilities and their corresponding countermeasures is conducted, focusing on aggregation algorithms that provide robustness against Byzantine failures. Following this direction, new aggregation algorithms are observed on a set of simulations that recreate realistic scenarios, in the absence and presence of Byzantine adversaries. These combine the Distance-based Krum approach with the Statistical based aggregation algorithm. Achieved results demonstrate the functionality of the proposed solutions in terms of accuracy and convergence rounds in comparison with well-known federated algorithms under a correct and incorrect estimation of the attackers number.","PeriodicalId":518411,"journal":{"name":"2024 IEEE 21st Consumer Communications & Networking Conference (CCNC)","volume":"9 8","pages":"1034-1035"},"PeriodicalIF":0.0,"publicationDate":"2024-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140531628","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-06DOI: 10.1109/CCNC51664.2024.10454737
Israt Ara, Brian Kelley
Physical Layer Security (PLS) as a native signaling enhancement to Layer-l security guarantees consumer privacy at the air interface. This paper proposes applying AIIML integration to Radio Access Networks (RAN) for enhanced 6G Internet of Things (loT) security. The paper defines a AIIML based PLS system model that can guarantee security against an eavesdropper. The paper also proposes an operational overview of AIIML integrated PLS with shared key agreement protocol in an O-RAN architecture for 6G IoT, and proposed sApps, that will lead the way of a new paradigm in low latency communication scheme for the consumers. Simulations of Key Bit Error rate (BER) detection in the presence of Rayleigh fading plus noise demonstrate that the AIIML-based PLS model jointly improves security and consumer communication performance.
{"title":"Natively Secure 6G IoT Using Intelligent Physical Layer Security","authors":"Israt Ara, Brian Kelley","doi":"10.1109/CCNC51664.2024.10454737","DOIUrl":"https://doi.org/10.1109/CCNC51664.2024.10454737","url":null,"abstract":"Physical Layer Security (PLS) as a native signaling enhancement to Layer-l security guarantees consumer privacy at the air interface. This paper proposes applying AIIML integration to Radio Access Networks (RAN) for enhanced 6G Internet of Things (loT) security. The paper defines a AIIML based PLS system model that can guarantee security against an eavesdropper. The paper also proposes an operational overview of AIIML integrated PLS with shared key agreement protocol in an O-RAN architecture for 6G IoT, and proposed sApps, that will lead the way of a new paradigm in low latency communication scheme for the consumers. Simulations of Key Bit Error rate (BER) detection in the presence of Rayleigh fading plus noise demonstrate that the AIIML-based PLS model jointly improves security and consumer communication performance.","PeriodicalId":518411,"journal":{"name":"2024 IEEE 21st Consumer Communications & Networking Conference (CCNC)","volume":"8 4","pages":"918-924"},"PeriodicalIF":0.0,"publicationDate":"2024-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140531629","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-06DOI: 10.1109/CCNC51664.2024.10454631
Simone Marasi, Stefano Ferretti
Money laundering in cryptocurrencies is a significant concern, as it facilitates and conceals crime and can distort markets and the broader financial system. To combat this issue, researchers have turned to techniques to develop effective Anti-Money Laundering (AML) frameworks. The findings contribute to the ongoing efforts to promote social good by reducing the impact of criminal activities on society. By preventing money laundering, we can also help to combat other criminal activities such as drug trafficking, corruption, and terrorism. This paper focuses on the use of Graph Neural Networks (GNNs) to classify cryptocurrencies transactions. Specifically, the study employs Graph Convolutional Networks (GCNs), Graph Attention Networks (GAT), the Chebyshev spatial convolutional neural network (ChebNet), and GraphSAGE network to classify Bitcoin transactions. The study finds that ChebNet, GraphSAGE and a variant of GAT outperform other methods and improve upon the state of the art in terms of recall and F1 scores, thus suggesting that they can be more reliable in identifying illicit transactions.
加密货币中的洗钱行为是一个令人严重关切的问题，因为它为犯罪提供了便利并掩盖了犯罪，可能会扭曲市场和更广泛的金融体系。为解决这一问题，研究人员已转向开发有效反洗钱（AML）框架的技术。这些研究成果有助于通过减少犯罪活动对社会的影响来促进社会公益的持续努力。通过防止洗钱，我们还可以帮助打击贩毒、腐败和恐怖主义等其他犯罪活动。本文重点研究使用图神经网络（GNN）对加密货币交易进行分类。具体来说，研究采用了图卷积网络（GCN）、图注意力网络（GAT）、切比雪夫空间卷积神经网络（ChebNet）和 GraphSAGE 网络来对比特币交易进行分类。研究发现，ChebNet、GraphSAGE 和 GAT 的一种变体在召回率和 F1 分数方面优于其他方法，并在技术水平上有所提高，从而表明它们在识别非法交易方面更加可靠。
{"title":"Anti-Money Laundering in Cryptocurrencies Through Graph Neural Networks: A Comparative Study","authors":"Simone Marasi, Stefano Ferretti","doi":"10.1109/CCNC51664.2024.10454631","DOIUrl":"https://doi.org/10.1109/CCNC51664.2024.10454631","url":null,"abstract":"Money laundering in cryptocurrencies is a significant concern, as it facilitates and conceals crime and can distort markets and the broader financial system. To combat this issue, researchers have turned to techniques to develop effective Anti-Money Laundering (AML) frameworks. The findings contribute to the ongoing efforts to promote social good by reducing the impact of criminal activities on society. By preventing money laundering, we can also help to combat other criminal activities such as drug trafficking, corruption, and terrorism. This paper focuses on the use of Graph Neural Networks (GNNs) to classify cryptocurrencies transactions. Specifically, the study employs Graph Convolutional Networks (GCNs), Graph Attention Networks (GAT), the Chebyshev spatial convolutional neural network (ChebNet), and GraphSAGE network to classify Bitcoin transactions. The study finds that ChebNet, GraphSAGE and a variant of GAT outperform other methods and improve upon the state of the art in terms of recall and F1 scores, thus suggesting that they can be more reliable in identifying illicit transactions.","PeriodicalId":518411,"journal":{"name":"2024 IEEE 21st Consumer Communications & Networking Conference (CCNC)","volume":"11 8","pages":"272-277"},"PeriodicalIF":0.0,"publicationDate":"2024-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140531797","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-06DOI: 10.1109/CCNC51664.2024.10454823
Ju Yeong Baek, Young-Seok Lee, Bang Chul Jung
We consider a downlink reconfigurable intelligent surfaces-based non-orthogonal multiple access technique with constellation adjustment (RIS-NOMA-CA) for 6G wireless communication systems. We mathematically analyze the bit-error-rate (BER) performance of the RIS-NOMA-CA technique assuming the optimal joint maximum likelihood (JML) detector at receivers. To the best of our knowledge, theoretical analysis of the downlink RIS-NOMA-CA technique has not been reported in the literature. We show that the RIS-NOMA-CA technique yields a better BER performance than the conventional RIS-NOMA technique without CA through computer simulations, and our analytical result matches well with simulation results.
{"title":"Downlink RIS-NOMA with Constellation Adjustment for 6G Wireless Communication Systems","authors":"Ju Yeong Baek, Young-Seok Lee, Bang Chul Jung","doi":"10.1109/CCNC51664.2024.10454823","DOIUrl":"https://doi.org/10.1109/CCNC51664.2024.10454823","url":null,"abstract":"We consider a downlink reconfigurable intelligent surfaces-based non-orthogonal multiple access technique with constellation adjustment (RIS-NOMA-CA) for 6G wireless communication systems. We mathematically analyze the bit-error-rate (BER) performance of the RIS-NOMA-CA technique assuming the optimal joint maximum likelihood (JML) detector at receivers. To the best of our knowledge, theoretical analysis of the downlink RIS-NOMA-CA technique has not been reported in the literature. We show that the RIS-NOMA-CA technique yields a better BER performance than the conventional RIS-NOMA technique without CA through computer simulations, and our analytical result matches well with simulation results.","PeriodicalId":518411,"journal":{"name":"2024 IEEE 21st Consumer Communications & Networking Conference (CCNC)","volume":"109 2","pages":"1076-1077"},"PeriodicalIF":0.0,"publicationDate":"2024-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140531799","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-06DOI: 10.1109/CCNC51664.2024.10454892
Christine Bassem
In crowdsensing platforms, algorithms and models for task allocation play a critical role in shaping user behaviors, engagement levels, the quality of the collected data, and the performance of the platform as a whole. Regardless of the sensing model, task allocation mechanisms are difficult to evaluate and benchmark. In contrast to evaluating deployments of crowd-sensing platforms with real crowds, they are often evaluated via simulators that are incapable of modeling the complexities of human behavior, specifically in terms of their commitment to the platform and quality of sensing, but their strength is the ability to rapidly experiment with multiple algorithms. In this paper, we abstract the general characteristics of participant behaviors in crowdsensing, and implement these characteristics within the TACSim simulation framework. Further exemplifying the extendability power of that simulation framework, and the benefits it can offer the crowdsensing community.
{"title":"Challenges of Modeling Participant Behavior in CrowdSensing Evaluation","authors":"Christine Bassem","doi":"10.1109/CCNC51664.2024.10454892","DOIUrl":"https://doi.org/10.1109/CCNC51664.2024.10454892","url":null,"abstract":"In crowdsensing platforms, algorithms and models for task allocation play a critical role in shaping user behaviors, engagement levels, the quality of the collected data, and the performance of the platform as a whole. Regardless of the sensing model, task allocation mechanisms are difficult to evaluate and benchmark. In contrast to evaluating deployments of crowd-sensing platforms with real crowds, they are often evaluated via simulators that are incapable of modeling the complexities of human behavior, specifically in terms of their commitment to the platform and quality of sensing, but their strength is the ability to rapidly experiment with multiple algorithms. In this paper, we abstract the general characteristics of participant behaviors in crowdsensing, and implement these characteristics within the TACSim simulation framework. Further exemplifying the extendability power of that simulation framework, and the benefits it can offer the crowdsensing community.","PeriodicalId":518411,"journal":{"name":"2024 IEEE 21st Consumer Communications & Networking Conference (CCNC)","volume":"106 4","pages":"1-6"},"PeriodicalIF":0.0,"publicationDate":"2024-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140531804","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-06DOI: 10.1109/CCNC51664.2024.10454773
Sudharshan Paindi Jayakumar, Alberto Conte
The growing demand for fast and reliable wireless services has led to the deployment of more base stations, which has made manual optimization of base station parameters more complex and time-consuming. This can lead to suboptimal network performance and a poor user experience. To address this challenge, we propose a Clustering-Driven Approach for Base Station Parameter Optimization and Automation (CeDA-BatOp), an automated framework for predicting optimized base station parameters. Our framework first compares three clustering algorithms: K-means, DBSCAN, and Agglomerative Clustering, selecting the most suitable one for specific scenarios based on their unique attributes. Simultaneously, our framework leverages machine learning (ML) algorithms to predict the optimal parameters for each base station with an evaluation of multiple ML models to identify the best fit for our data. It also incorporates data drift monitoring to track gradual changes in data distribution over time, ensuring ML model accuracy through periodic retraining. In the simulated scenario, our framework achieved an average of 76% reduction in memory overhead and simplified training by utilizing fewer models, effectively minimizing computational resources. The drift detection system demonstrated an exceptional accuracy of 98.87%, outperforming other cases. These results highlight the potential of our framework to significantly benefit network operators by automating base station parameter tuning, reducing human involvement, and substantially improving network performance and cost savings.
对快速、可靠的无线服务日益增长的需求导致基站数量的增加，从而使基站参数的人工优化变得更加复杂和耗时。这可能导致网络性能不达标和用户体验不佳。为了应对这一挑战，我们提出了基站参数优化和自动化的聚类驱动方法（CeDA-BatOp），这是一个预测优化基站参数的自动化框架。我们的框架首先比较了三种聚类算法：我们的框架首先比较了三种聚类算法：K-means、DBSCAN 和聚合聚类，并根据其独特属性为特定场景选择最合适的算法。同时，我们的框架利用机器学习（ML）算法预测每个基站的最佳参数，并对多个 ML 模型进行评估，以确定最适合我们数据的模型。它还结合了数据漂移监测，以跟踪数据分布随时间的逐渐变化，通过定期再训练确保 ML 模型的准确性。在模拟场景中，我们的框架平均减少了 76% 的内存开销，并通过使用更少的模型简化了训练，有效地最大限度地减少了计算资源。漂移检测系统的准确率高达 98.87%，优于其他案例。这些结果凸显了我们的框架的潜力，它通过自动调整基站参数、减少人工参与、大幅提高网络性能和节约成本，使网络运营商受益匪浅。
{"title":"Framework: Clustering-Driven Approach for Base Station Parameter Optimization and Automation (CeDA-BatOp)","authors":"Sudharshan Paindi Jayakumar, Alberto Conte","doi":"10.1109/CCNC51664.2024.10454773","DOIUrl":"https://doi.org/10.1109/CCNC51664.2024.10454773","url":null,"abstract":"The growing demand for fast and reliable wireless services has led to the deployment of more base stations, which has made manual optimization of base station parameters more complex and time-consuming. This can lead to suboptimal network performance and a poor user experience. To address this challenge, we propose a Clustering-Driven Approach for Base Station Parameter Optimization and Automation (CeDA-BatOp), an automated framework for predicting optimized base station parameters. Our framework first compares three clustering algorithms: K-means, DBSCAN, and Agglomerative Clustering, selecting the most suitable one for specific scenarios based on their unique attributes. Simultaneously, our framework leverages machine learning (ML) algorithms to predict the optimal parameters for each base station with an evaluation of multiple ML models to identify the best fit for our data. It also incorporates data drift monitoring to track gradual changes in data distribution over time, ensuring ML model accuracy through periodic retraining. In the simulated scenario, our framework achieved an average of 76% reduction in memory overhead and simplified training by utilizing fewer models, effectively minimizing computational resources. The drift detection system demonstrated an exceptional accuracy of 98.87%, outperforming other cases. These results highlight the potential of our framework to significantly benefit network operators by automating base station parameter tuning, reducing human involvement, and substantially improving network performance and cost savings.","PeriodicalId":518411,"journal":{"name":"2024 IEEE 21st Consumer Communications & Networking Conference (CCNC)","volume":"102 7","pages":"1026-1029"},"PeriodicalIF":0.0,"publicationDate":"2024-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140531813","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-06DOI: 10.1109/CCNC51664.2024.10454636
Laura Belli, Luca Davoli, Gianluigi Ferrari
Localization services for precise and continuous monitoring of the locations of both humans and vehicles in industrial environments are among the most relevant applications in Industrial Internet of Things (IIoT) contexts, to maximize safety and optimize operational activities. Unfortunately, localization in industrial scenarios is particularly challenging because targets can generally move freely in both indoor and outdoor areas. In this paper, we propose a localization monitoring architecture based on a prototypical wearable IoT device equipped with Ultra-Wide Band (UWB), inertial, and GNSS/RTK technologies for seamless localization in heterogeneous environments. We focus on a Web of Things (WoT) approach, verifying suitability and limitations in a real use case scenario. Our approach shows that the proposed architecture can effectively enhance the safety of workers, detecting potentially dangerous events and triggering alarms (e.g., via smart buzzers or gas concentration warning devices) based on a cloud WoT architecture.
{"title":"A Cloud-Oriented Indoor-Outdoor Real-Time Localization IoT Architecture for Industrial Environments","authors":"Laura Belli, Luca Davoli, Gianluigi Ferrari","doi":"10.1109/CCNC51664.2024.10454636","DOIUrl":"https://doi.org/10.1109/CCNC51664.2024.10454636","url":null,"abstract":"Localization services for precise and continuous monitoring of the locations of both humans and vehicles in industrial environments are among the most relevant applications in Industrial Internet of Things (IIoT) contexts, to maximize safety and optimize operational activities. Unfortunately, localization in industrial scenarios is particularly challenging because targets can generally move freely in both indoor and outdoor areas. In this paper, we propose a localization monitoring architecture based on a prototypical wearable IoT device equipped with Ultra-Wide Band (UWB), inertial, and GNSS/RTK technologies for seamless localization in heterogeneous environments. We focus on a Web of Things (WoT) approach, verifying suitability and limitations in a real use case scenario. Our approach shows that the proposed architecture can effectively enhance the safety of workers, detecting potentially dangerous events and triggering alarms (e.g., via smart buzzers or gas concentration warning devices) based on a cloud WoT architecture.","PeriodicalId":518411,"journal":{"name":"2024 IEEE 21st Consumer Communications & Networking Conference (CCNC)","volume":"64 1","pages":"1-6"},"PeriodicalIF":0.0,"publicationDate":"2024-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140531829","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-06DOI: 10.1109/CCNC51664.2024.10454778
Abiola Adegboyega
The cloud hosts diverse applications with different workload characteristics. Public cloud traces provide opportunities for analysis to gain insights informing autoscaling, forecasting among other operations. This paper presents the statistical analysis of a recent Alibaba cloud storage workload. The isolation & aggregation of all read/write time-series per recorded workload was done. Application of statistical methods yielded novel distributions from which forecasting solutions integrating time-varying variance captured workload burstiness. A 25% improvement in forecasting accuracy over current methods was achieved. The set of workload time-series has been made available online for further analysis by the research community.
{"title":"Cloud Storage Workload Characterization: An Approach with Time-Series Analysis","authors":"Abiola Adegboyega","doi":"10.1109/CCNC51664.2024.10454778","DOIUrl":"https://doi.org/10.1109/CCNC51664.2024.10454778","url":null,"abstract":"The cloud hosts diverse applications with different workload characteristics. Public cloud traces provide opportunities for analysis to gain insights informing autoscaling, forecasting among other operations. This paper presents the statistical analysis of a recent Alibaba cloud storage workload. The isolation & aggregation of all read/write time-series per recorded workload was done. Application of statistical methods yielded novel distributions from which forecasting solutions integrating time-varying variance captured workload burstiness. A 25% improvement in forecasting accuracy over current methods was achieved. The set of workload time-series has been made available online for further analysis by the research community.","PeriodicalId":518411,"journal":{"name":"2024 IEEE 21st Consumer Communications & Networking Conference (CCNC)","volume":"43 9","pages":"1090-1091"},"PeriodicalIF":0.0,"publicationDate":"2024-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140531860","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-06DOI: 10.1109/CCNC51664.2024.10454680
Giovanni Delnevo, Manuel Andruccioli, S. Mirri
The widespread diffusion of Large Language Models (LLMs) has ushered in a transformative era across numerous research domains, including web accessibility. In fact, they can potentially offer automated solutions for generating accessible content, performing accessibility testing, and enhancing the overall user experience for individuals with disabilities. In this paper, we investigate how LLMs can be successfully employed to evaluate and correct web accessibility. Then, we delve into the positive implications and the current challenges derived from the interaction between developers and LLMs in this specific context. Finally, we present some future directions that could be explored to ensure that web content remains accessible to all.
{"title":"On the Interaction with Large Language Models for Web Accessibility: Implications and Challenges","authors":"Giovanni Delnevo, Manuel Andruccioli, S. Mirri","doi":"10.1109/CCNC51664.2024.10454680","DOIUrl":"https://doi.org/10.1109/CCNC51664.2024.10454680","url":null,"abstract":"The widespread diffusion of Large Language Models (LLMs) has ushered in a transformative era across numerous research domains, including web accessibility. In fact, they can potentially offer automated solutions for generating accessible content, performing accessibility testing, and enhancing the overall user experience for individuals with disabilities. In this paper, we investigate how LLMs can be successfully employed to evaluate and correct web accessibility. Then, we delve into the positive implications and the current challenges derived from the interaction between developers and LLMs in this specific context. Finally, we present some future directions that could be explored to ensure that web content remains accessible to all.","PeriodicalId":518411,"journal":{"name":"2024 IEEE 21st Consumer Communications & Networking Conference (CCNC)","volume":"83 3","pages":"1-6"},"PeriodicalIF":0.0,"publicationDate":"2024-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140531888","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-06DOI: 10.1109/CCNC51664.2024.10454875
Rukayat Olapojoye, Mohamed Baza, Tara Salman
Undoubtedly, Machine Learning (ML) has revolutionized many applications in recent years. A vast amount of heterogeneous data distributed globally is being used to build efficient and robust prediction models. This has led to the need for decentralized ML paradigms. Federated Learning (FL) has emerged as a decentralized ML paradigm that creates global models from multiple privately trained local datasets. Nevertheless, FL comes with some challenges, such as using a central server, leading to a single point of failure and trust issues. Blockchain-based Federated learning (BFL) has been proposed to resolve these challenges. However, due to the openness of the Blockchain system, malicious clients can access critical information, such as the number of participating clients, and launch attacks on the BFL system. This paper presents a practicable model poisoning attack on BFL systems. Several experiments are conducted with different attack scenarios and settings explored. The evaluations and results show the efficacy and impact of the model poisoning.
毫无疑问，机器学习（ML）近年来给许多应用带来了变革。分布在全球的大量异构数据被用来建立高效、稳健的预测模型。这导致了对分散式 ML 模式的需求。联邦学习（Federated Learning，FL）作为一种去中心化的 ML 范式已经出现，它能从多个私下训练的本地数据集创建全局模型。不过，FL 也面临一些挑战，例如使用中央服务器会导致单点故障和信任问题。为了解决这些难题，有人提出了基于区块链的联合学习（BFL）。然而，由于区块链系统的开放性，恶意客户可以获取关键信息，如参与客户的数量，并对 BFL 系统发起攻击。本文提出了一种针对 BFL 系统的实用模型中毒攻击。本文针对不同的攻击场景和设置进行了多次实验。评估和结果表明了模型中毒的功效和影响。
{"title":"On the Analysis of Model Poisoning Attacks Against Blockchain-Based Federated Learning","authors":"Rukayat Olapojoye, Mohamed Baza, Tara Salman","doi":"10.1109/CCNC51664.2024.10454875","DOIUrl":"https://doi.org/10.1109/CCNC51664.2024.10454875","url":null,"abstract":"Undoubtedly, Machine Learning (ML) has revolutionized many applications in recent years. A vast amount of heterogeneous data distributed globally is being used to build efficient and robust prediction models. This has led to the need for decentralized ML paradigms. Federated Learning (FL) has emerged as a decentralized ML paradigm that creates global models from multiple privately trained local datasets. Nevertheless, FL comes with some challenges, such as using a central server, leading to a single point of failure and trust issues. Blockchain-based Federated learning (BFL) has been proposed to resolve these challenges. However, due to the openness of the Blockchain system, malicious clients can access critical information, such as the number of participating clients, and launch attacks on the BFL system. This paper presents a practicable model poisoning attack on BFL systems. Several experiments are conducted with different attack scenarios and settings explored. The evaluations and results show the efficacy and impact of the model poisoning.","PeriodicalId":518411,"journal":{"name":"2024 IEEE 21st Consumer Communications & Networking Conference (CCNC)","volume":"71 3","pages":"943-949"},"PeriodicalIF":0.0,"publicationDate":"2024-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140531923","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：