This is an autogenerated issue. It uses the facebook/bart-large-cnn model to generate summaries for each papers appeared below.
Keyword: IoT (14 papers)
Differentiating Viral and Bacterial Infections: A Machine Learning Model Based on Routine Blood Test Values
:brain: Authors: Gregor Gunčar, Matjaž Kukar, Tim Smole, Sašo Moškon, Tomaž Vovko, Simon Podnar, Peter Černelč, Miran Brvar, Mateja Notar, Manca Köster, Marjeta Tušek Jelenc, Marko Notar :paw_prints: Details:arXiv:2305.07877 :ramen: tl;dr: The growing threat of antibiotic resistance necessitates accurate differentiation between bacterial and viral infections for proper antibiotic administration. A Virus vs. Bacteria machine learning model was developed to discern between these infection types using 16 routine blood test results, C-reactive protein levels, biological sex, and age. With a dataset of 44,120 cases from a single medical center, the model demonstrated remarkable accuracy of 82.2%.... :seedling: Read more →
:brain: Authors: Xiaofei Guan, Boya Hu, Shipeng Mao, Xintong Wang :paw_prints: Details:arXiv:2305.07940 :ramen: tl;dr: Modeling and control of the magnetohydrodynamics (MHD) system remain a challenging problem. We develop the MHDnet as a physics-informed learning approach to MHD problems. Multi-modes multiscale feature embedding into multiscales neural network architecture can accelerate the convergence of the neural networks.... :seedling: Read more →
On the Capacity of DNA Labeling
:brain: Authors: Dganit Hanania, Daniella Bar-Lev, Yevgeni Nogin, Yoav Shechtman, Eitan Yaakobi :paw_prints: Details:arXiv:2305.07992 :ramen: tl;dr: DNA labeling is a powerful tool in molecular biology and biotechnology that allows for the visualization, detection, and study of DNA at the molecular level. Under this paradigm, a DNA molecule is being labeled by specific k patterns and is then imaged. The labeling capacity is computed for any single label and several results are provided for multiple labels as well.... :seedling: Read more →
Activity-based Profiling for Energy Harvesting Estimation
:brain: Authors: Jiajie Li, Amani Abusafia, Abdallah Lakhdari, Athman Bouguettaya :paw_prints: Details:arXiv:2305.08025 :ramen: tl;dr: We propose a novel activity-based profiling framework to estimate IoT users' harvested energy based on their daily activities. Energy is harvested from natural sources such as the kinetic movement of IoT users. The profiling framework captures the users' physical activity data to define activity- based profiles. These profiles are utilized to estimate the harvested energy by IoT user.... :sunflower: Comments: 2 pages, 2 Figures,This is an accepted posterpaper at the ACM/IEEE International Conference on Information Processing in Sensor Networks (IPSN) 2023 :seedling: Read more →
Privacy-preserving Blockchain-enabled Parametric Insurance via Remote Sensing and IoT
:brain: Authors: Mingyu Hao, Keyang Qian, Sid Chi-Kin Chau :paw_prints: Details:arXiv:2305.08384 :ramen: tl;dr: Traditional insurance has suffered from high operational costs, opaqueness, inefficiency and a lack of trust. "parametric insurance" aims to overcome these issues by automating the underwriting and claim processes of insurance policies on a blockchain. The openness of blockchain platforms raises a concern of user privacy, as the private user data in insurance claims may be exposed.... :seedling: Read more →
Marsellus: A Heterogeneous RISC-V AI-IoT End-Node SoC with 2-to-8b DNN Acceleration and 30%-Boost Adaptive Body Biasing
:brain: Authors: Francesco Conti, Gianna Paulin, Davide Rossi, Alfio Di Mauro, Georg Rutishauser, Gianmarco Ottavi, Manuel Eggimann, Hayate Okuhara, Luca Benini :paw_prints: Details:arXiv:2305.08415 :ramen: tl;dr: Artificial Intelligence-enabled Internet-of-Things (AI-IoT) System-on-a-Chip (SoC) for augmented reality, personalized healthcare, and nano-robotics need to run many diverse tasks within a power envelope of a few tens of mW. We present Marsellus, an all-digital heterogeneous SoC for AI-T end-nodes fabricated in GlobalFoundries 22nm FDX.... :sunflower: Comments: Pre-print submitted to IEEE Journal of Solid-State Circuits :seedling: Read more →
FeatFSDA: Towards Few-shot Domain Adaptation for Video-based Activity Recognition
:brain: Authors: Kunyu Peng, Di Wen, David Schneider, Jiaming Zhang, Kailun Yang, M. Saquib Sarfraz, Rainer Stiefelhagen, Alina Roitberg :paw_prints: Details:arXiv:2305.08420 :ramen: tl;dr: FSDA-AR leverages a very small amount of labeled target videos to achieve effective adaptation. FeatFSDA incorporates a latent space semantic adjacency loss, a domain prototypical similarity loss, and a graph-attentive-network-based edge dropout technique.... :sunflower: Comments: Benchmarks and code will be released at this https URL :seedling: Read more →
Task-Oriented Communication Design at Scale
:brain: Authors: Arsham Mostaani, Thang X. Vu, Hamed Habibi, Symeon Chatzinotas, Bjorn Ottersten :paw_prints: Details:arXiv:2305.08481 :ramen: tl;dr: Task-oriented communication design (TOCD) is getting accelerated attention from the research community. The computational complexity of learning the VoI grows exponentially with the number of agents. We observe that one can reduce the computational cost of obtaining the value of information by exploiting insights gained from studying a similar two-agent system.... :seedling: Read more →
FLARE: Detection and Mitigation of Concept Drift for Federated Learning based IoT Deployments
:brain: Authors: Theo Chow, Usman Raza, Ioannis Mavromatis, Aftab Khan :paw_prints: Details:arXiv:2305.08504 :ramen: tl;dr: In traditional cloud-centric approaches, raw data is transmitted to a central server for training and inference purposes. Federated Learning migrates both tasks closer to the edge nodes and endpoints. This allows for a significant reduction in data exchange while preserving the privacy of users. Trained models, though, may under-perform in dynamic environments due to changes in the data distribution.... :sunflower: Comments: To appear at IWCMC 2023, Marrakesh, Morocco :seedling: Read more →
Benchmarking UWB-Based Infrastructure-Free Positioning and Multi-Robot Relative Localization: Dataset and Characterization
:brain: Authors: Paola Torrico Morón, Sahar Salimpour, Lei Fu, Xianjia Yu, Jorge Peña Queralta, Tomi Westerlund :paw_prints: Details:arXiv:2305.08532 :ramen: tl;dr: Ultra-wideband (UWB) positioning has emerged as a low-cost and dependable localization solution for multiple use cases, from mobile robots to asset tracking within the Industrial IoT. Research in UWB-based relative localization and infrastructure-free localization is gaining traction, further domains. tools and datasets in this domain are scarce.... :sunflower: Comments: 6 pages, 8 figures, 3 tables :seedling: Read more →
Fast Inference of Tree Ensembles on ARM Devices
:brain: Authors: Simon Koschel, Sebastian Buschjäger, Claudio Lucchese, Katharina Morik :paw_prints: Details:arXiv:2305.08579 :ramen: tl;dr: A careful implementation of tree traversal on ARM CPUs leads to a speed-up of up to 9.4 compared to a reference implementation. quantized models seem to outperform models using floating-point values in terms of speed in almost all cases, with a neglectable impact on the predictive performance of the model.... :sunflower: Comments: 12 pages, 2 figures, 4 algorithms :seedling: Read more →
A Survey of Blockchain and Artificial Intelligence for 6G Wireless Communications
:brain: Authors: Yiping Zuo, Jiajia Guo, Ning Gao, Yongxu Zhu, Shi Jin, Xiao Li :paw_prints: Details:arXiv:2305.08604 :ramen: tl;dr: The research on the sixth-generation (6G) wireless communications for the development of future mobile communication networks has been officially launched around the world. 6G networks face multifarious challenges, such as resource-constrained mobile devices, difficult wireless resource management, high complexity of heterogeneous network architectures, explosive computing and storage requirements, privacy and security threats. To address these challenges, deploying blockchain and artificial intelligence (AI) in 6G Networks may realize new breakthroughs in advancing network performances.... :seedling: Read more →
Quantum-Based Combinatorial Optimization for Structural Health Monitoring Applications
:brain: Authors: Gabriel San Martin Silva, Enrique Lopez Droguett :paw_prints: Details:arXiv:2305.08738 :ramen: tl;dr: Structural Health Monitoring (SHM) arises as an especially interesting field to integrate and develop these new sensing capabilities. One of the main challenges when designing a modern monitoring system is the Optimal Sensor Placement (OSP) problem. This study proposes a novel quantum-based combinatorial optimization approach to solve the OSP problem.... :seedling: Read more →
Automating privacy decisions -- where to draw the line?
:brain: Authors: Victor Morel, Simone Fischer-Hübner :paw_prints: Details:arXiv:2305.08747 :ramen: tl;dr: Users are often overwhelmed by privacy decisions to manage their personal data. These decisions can take various forms, and each can have different legal impacts. Scholars and industry have been proposing tools to better automate the process of privacy decisions at different levels. We provide an overview of the main challenges raised by the automation of Privacy decisions.... :sunflower: Comments: 9 pages, 2 figures, accepted for publication at iwpe.info 2023 :seedling: Read more →
Keyword: edge computing (2 papers)
MetaMorphosis: Task-oriented Privacy Cognizant Feature Generation for Multi-task Learning
:brain: Authors: Md Adnan Arefeen, Zhouyu Li, Md Yusuf Sarwar Uddin, Anupam Das :paw_prints: Details:arXiv:2305.07815 :ramen: tl;dr: Running separate single-task models on edge devices is inefficient regarding the required computational resource and time. This paper proposes a novel deep learning-based privacy-cognizant feature generation process called MetaMorphosis. It outperforms recent adversarial learning and universal feature generation methods by guaranteeing privacy requirements.... :sunflower: Comments: Preprint version, 22 pages. Keywords: Multi-task learning, neural networks, collaborative intelligence, differential privacy, task privacy :seedling: Read more →
Attention-based QoE-aware Digital Twin Empowered Edge Computing for Immersive Virtual Reality
:brain: Authors: Jiadong Yu, Ahmad Alhilal, Tailin Zhou, Hui Pan, Danny H.K. Tsang :paw_prints: Details:arXiv:2305.08569 :ramen: tl;dr: Metaverse applications such as virtual reality (VR) content streaming, require optimal resource allocation strategies for mobile edge computing (MEC) In contrast to online reinforcement learning (RL) algorithms, which can incur substantial communication overheads and longer delays, the majority of existing works employ offline-trained RL algorithms for resource allocation decisions.... :seedling: Read more →
Keyword: Blockchain (1 papers)
Interchain Timestamping for Mesh Security
:brain: Authors: Ertem Nusret Tas, Runchao Han, David Tse, Fisher Yu, Kamilla Nazirkhanova :paw_prints: Details:arXiv:2305.07830 :ramen: tl;dr: Interchain timestamping can be readily implemented for Cosmos chains connected by the Inter-Blockchain Communication (IBC) protocol. We show that it extracts the maximum economic security from the provider chains, as quantified by the slashable safety resilience. We compare interchain timestamps with cross-staking, the original solution to mesh security, and with Trustboost.... :sunflower: Comments: Forthcoming in ACM Conference on Computer and Communications Security (CCS) 2023 :seedling: Read more →
Keyword: Federated Learning (8 papers)
Understanding Model Averaging in Federated Learning on Heterogeneous Data
:brain: Authors: Tailin Zhou, Zehong Lin, Jun Zhang, Danny H.K. Tsang :paw_prints: Details:arXiv:2305.07845 :ramen: tl;dr: Model averaging is a widely adopted technique in federated learning (FL) It aggregates multiple client models trained on heterogeneous data to obtain a well-performed global model. We decompose the expected prediction error of the global model into five factors related to client models. We propose adopting iterative moving averaging (IMA) on global models.... :seedling: Read more →
A Federated Learning-based Industrial Health Prognostics for Heterogeneous Edge Devices using Matched Feature Extraction
:brain: Authors: Anushiya Arunan, Yan Qin, Xiaoli Li, Chau Yuen :paw_prints: Details:arXiv:2305.07854 :ramen: tl;dr: Industrial health prognostics require rich training data to develop accurate and reliable predictive models. The abundance of edge industrial data necessitate decentralized data utilization. We propose a pioneering FL-based health prognostic model with a feature similarity-matched parameter aggregation algorithm to discriminately learn from heterogeneous edge data.... :sunflower: Comments: This paper has been accepted by IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING :seedling: Read more →
A Survey of Federated Evaluation in Federated Learning
:brain: Authors: Behnaz Soltani, Yipeng Zhou, Venus Haghighi, John C.S. Lui :paw_prints: Details:arXiv:2305.08070 :ramen: tl;dr: In traditional machine learning, it is trivial to conduct model evaluation since all data samples are managed centrally by a server. In federated learning (FL), clients do not expose their original data to preserve data privacy. Federated evaluation plays a vital role in client selection, incentive mechanism design, malicious attack detection, etc.... :seedling: Read more →
Federated TD Learning over Finite-Rate Erasure Channels: Linear Speedup under Markovian Sampling
:brain: Authors: Nicolò Dal Fabbro, Aritra Mitra, George J. Pappas :paw_prints: Details:arXiv:2305.08104 :ramen: tl;dr: Federated learning (FL) has recently gained much attention due to its effectiveness in speeding up supervised learning tasks. However, whether similar speedups can be established for reinforcement learning remains much less understood theoretically. We propose and analyze QFedTD - a quantized federated temporal difference learning algorithm with linear function approximation.... :seedling: Read more →
Balancing Privacy and Utility of Spatio-Temporal Data for Taxi-Demand Prediction
:brain: Authors: Yumeki Goto, Tomoya Matsumoto, Hamada Rizk, Naoto Yanai, Hirozumi Yamaguchi :paw_prints: Details:arXiv:2305.08107 :ramen: tl;dr: Taxi-demand prediction is an important application of machine learning. The use of sensitive data in these systems raises concerns about privacy and security. We propose a system that utilizes region-independent encoding for geographic lat-long coordinates. We employ cost-sensitive learning and various regularization techniques to mitigate issues related to data scarcity and overfitting.... :seedling: Read more →
FedAds: A Benchmark for Privacy-Preserving CVR Estimation with Vertical Federated Learning
:brain: Authors: Penghui Wei, Hongjian Dou, Shaoguo Liu, Rongjun Tang, Li Liu, Liang Wang, Bo Zheng :paw_prints: Details:arXiv:2305.08328 :ramen: tl;dr: Conversion rate (CVR) estimation aims to predict the probability of conversion event after a user has clicked an ad. To estimate CVR accurately and protect data privacy better, vertical federated learning (vFL) is a natural solution. FedAds is the first benchmark for CVR estimation with vFL.... :sunflower: Comments: SIGIR 2023, Resource Track :seedling: Read more →
Quadratic Functional Encryption for Secure Training in Vertical Federated Learning
:brain: Authors: Shuangyi Chen, Anuja Modi, Shweta Agrawal, Ashish Khisti :paw_prints: Details:arXiv:2305.08358 :ramen: tl;dr: Vertical federated learning (VFL) enables the collaborative training of machine learning (ML) models. In VFL, the labels are available to a single party and the complete feature set is formed only when data from all parties is combined. Xu et al. proposed a new framework called FedV for secure gradient computation for VFL.... :sunflower: Comments: Accepted by ISIT 2023 :seedling: Read more →
ESAFL: Efficient Secure Additively Homomorphic Encryption for Cross-Silo Federated Learning
:brain: Authors: Jiahui Wu, Weizhe Zhang, Fucai Luo :paw_prints: Details:arXiv:2305.08599 :ramen: tl;dr: Cross-silo federated learning (FL) enables multiple clients to collaboratively train a machine learning model without sharing training data. Techniques using homomorphic encryption (HE) have been designed to solve this but bring their own challenges. We present a new FL protocol ESAFL that leverages a novel efficient and secure additively HE.... :seedling: Read more →
This is an autogenerated issue. It uses the facebook/bart-large-cnn model to generate summaries for each papers appeared below.
Keyword: IoT (14 papers)
Differentiating Viral and Bacterial Infections: A Machine Learning Model Based on Routine Blood Test Values
:brain: Authors: Gregor Gunčar, Matjaž Kukar, Tim Smole, Sašo Moškon, Tomaž Vovko, Simon Podnar, Peter Černelč, Miran Brvar, Mateja Notar, Manca Köster, Marjeta Tušek Jelenc, Marko Notar:paw_prints: Details: arXiv:2305.07877
:ramen: tl;dr: The growing threat of antibiotic resistance necessitates accurate differentiation between bacterial and viral infections for proper antibiotic administration. A Virus vs. Bacteria machine learning model was developed to discern between these infection types using 16 routine blood test results, C-reactive protein levels, biological sex, and age. With a dataset of 44,120 cases from a single medical center, the model demonstrated remarkable accuracy of 82.2%....
:seedling: Read more →
MHDnet: Multi-modes multiscale physics informed neural networks for solving Magnetohydrodynamics problems
:brain: Authors: Xiaofei Guan, Boya Hu, Shipeng Mao, Xintong Wang:paw_prints: Details: arXiv:2305.07940
:ramen: tl;dr: Modeling and control of the magnetohydrodynamics (MHD) system remain a challenging problem. We develop the MHDnet as a physics-informed learning approach to MHD problems. Multi-modes multiscale feature embedding into multiscales neural network architecture can accelerate the convergence of the neural networks....
:seedling: Read more →
On the Capacity of DNA Labeling
:brain: Authors: Dganit Hanania, Daniella Bar-Lev, Yevgeni Nogin, Yoav Shechtman, Eitan Yaakobi:paw_prints: Details: arXiv:2305.07992
:ramen: tl;dr: DNA labeling is a powerful tool in molecular biology and biotechnology that allows for the visualization, detection, and study of DNA at the molecular level. Under this paradigm, a DNA molecule is being labeled by specific k patterns and is then imaged. The labeling capacity is computed for any single label and several results are provided for multiple labels as well....
:seedling: Read more →
Activity-based Profiling for Energy Harvesting Estimation
:brain: Authors: Jiajie Li, Amani Abusafia, Abdallah Lakhdari, Athman Bouguettaya:paw_prints: Details: arXiv:2305.08025
:ramen: tl;dr: We propose a novel activity-based profiling framework to estimate IoT users' harvested energy based on their daily activities. Energy is harvested from natural sources such as the kinetic movement of IoT users. The profiling framework captures the users' physical activity data to define activity- based profiles. These profiles are utilized to estimate the harvested energy by IoT user....
:sunflower: Comments: 2 pages, 2 Figures,This is an accepted posterpaper at the ACM/IEEE International Conference on Information Processing in Sensor Networks (IPSN) 2023
:seedling: Read more →
Privacy-preserving Blockchain-enabled Parametric Insurance via Remote Sensing and IoT
:brain: Authors: Mingyu Hao, Keyang Qian, Sid Chi-Kin Chau:paw_prints: Details: arXiv:2305.08384
:ramen: tl;dr: Traditional insurance has suffered from high operational costs, opaqueness, inefficiency and a lack of trust. "parametric insurance" aims to overcome these issues by automating the underwriting and claim processes of insurance policies on a blockchain. The openness of blockchain platforms raises a concern of user privacy, as the private user data in insurance claims may be exposed....
:seedling: Read more →
Marsellus: A Heterogeneous RISC-V AI-IoT End-Node SoC with 2-to-8b DNN Acceleration and 30%-Boost Adaptive Body Biasing
:brain: Authors: Francesco Conti, Gianna Paulin, Davide Rossi, Alfio Di Mauro, Georg Rutishauser, Gianmarco Ottavi, Manuel Eggimann, Hayate Okuhara, Luca Benini:paw_prints: Details: arXiv:2305.08415
:ramen: tl;dr: Artificial Intelligence-enabled Internet-of-Things (AI-IoT) System-on-a-Chip (SoC) for augmented reality, personalized healthcare, and nano-robotics need to run many diverse tasks within a power envelope of a few tens of mW. We present Marsellus, an all-digital heterogeneous SoC for AI-T end-nodes fabricated in GlobalFoundries 22nm FDX....
:sunflower: Comments: Pre-print submitted to IEEE Journal of Solid-State Circuits
:seedling: Read more →
FeatFSDA: Towards Few-shot Domain Adaptation for Video-based Activity Recognition
:brain: Authors: Kunyu Peng, Di Wen, David Schneider, Jiaming Zhang, Kailun Yang, M. Saquib Sarfraz, Rainer Stiefelhagen, Alina Roitberg:paw_prints: Details: arXiv:2305.08420
:ramen: tl;dr: FSDA-AR leverages a very small amount of labeled target videos to achieve effective adaptation. FeatFSDA incorporates a latent space semantic adjacency loss, a domain prototypical similarity loss, and a graph-attentive-network-based edge dropout technique....
:sunflower: Comments: Benchmarks and code will be released at this https URL
:seedling: Read more →
Task-Oriented Communication Design at Scale
:brain: Authors: Arsham Mostaani, Thang X. Vu, Hamed Habibi, Symeon Chatzinotas, Bjorn Ottersten:paw_prints: Details: arXiv:2305.08481
:ramen: tl;dr: Task-oriented communication design (TOCD) is getting accelerated attention from the research community. The computational complexity of learning the VoI grows exponentially with the number of agents. We observe that one can reduce the computational cost of obtaining the value of information by exploiting insights gained from studying a similar two-agent system....
:seedling: Read more →
FLARE: Detection and Mitigation of Concept Drift for Federated Learning based IoT Deployments
:brain: Authors: Theo Chow, Usman Raza, Ioannis Mavromatis, Aftab Khan:paw_prints: Details: arXiv:2305.08504
:ramen: tl;dr: In traditional cloud-centric approaches, raw data is transmitted to a central server for training and inference purposes. Federated Learning migrates both tasks closer to the edge nodes and endpoints. This allows for a significant reduction in data exchange while preserving the privacy of users. Trained models, though, may under-perform in dynamic environments due to changes in the data distribution....
:sunflower: Comments: To appear at IWCMC 2023, Marrakesh, Morocco
:seedling: Read more →
Benchmarking UWB-Based Infrastructure-Free Positioning and Multi-Robot Relative Localization: Dataset and Characterization
:brain: Authors: Paola Torrico Morón, Sahar Salimpour, Lei Fu, Xianjia Yu, Jorge Peña Queralta, Tomi Westerlund:paw_prints: Details: arXiv:2305.08532
:ramen: tl;dr: Ultra-wideband (UWB) positioning has emerged as a low-cost and dependable localization solution for multiple use cases, from mobile robots to asset tracking within the Industrial IoT. Research in UWB-based relative localization and infrastructure-free localization is gaining traction, further domains. tools and datasets in this domain are scarce....
:sunflower: Comments: 6 pages, 8 figures, 3 tables
:seedling: Read more →
Fast Inference of Tree Ensembles on ARM Devices
:brain: Authors: Simon Koschel, Sebastian Buschjäger, Claudio Lucchese, Katharina Morik:paw_prints: Details: arXiv:2305.08579
:ramen: tl;dr: A careful implementation of tree traversal on ARM CPUs leads to a speed-up of up to 9.4 compared to a reference implementation. quantized models seem to outperform models using floating-point values in terms of speed in almost all cases, with a neglectable impact on the predictive performance of the model....
:sunflower: Comments: 12 pages, 2 figures, 4 algorithms
:seedling: Read more →
A Survey of Blockchain and Artificial Intelligence for 6G Wireless Communications
:brain: Authors: Yiping Zuo, Jiajia Guo, Ning Gao, Yongxu Zhu, Shi Jin, Xiao Li:paw_prints: Details: arXiv:2305.08604
:ramen: tl;dr: The research on the sixth-generation (6G) wireless communications for the development of future mobile communication networks has been officially launched around the world. 6G networks face multifarious challenges, such as resource-constrained mobile devices, difficult wireless resource management, high complexity of heterogeneous network architectures, explosive computing and storage requirements, privacy and security threats. To address these challenges, deploying blockchain and artificial intelligence (AI) in 6G Networks may realize new breakthroughs in advancing network performances....
:seedling: Read more →
Quantum-Based Combinatorial Optimization for Structural Health Monitoring Applications
:brain: Authors: Gabriel San Martin Silva, Enrique Lopez Droguett:paw_prints: Details: arXiv:2305.08738
:ramen: tl;dr: Structural Health Monitoring (SHM) arises as an especially interesting field to integrate and develop these new sensing capabilities. One of the main challenges when designing a modern monitoring system is the Optimal Sensor Placement (OSP) problem. This study proposes a novel quantum-based combinatorial optimization approach to solve the OSP problem....
:seedling: Read more →
Automating privacy decisions -- where to draw the line?
:brain: Authors: Victor Morel, Simone Fischer-Hübner:paw_prints: Details: arXiv:2305.08747
:ramen: tl;dr: Users are often overwhelmed by privacy decisions to manage their personal data. These decisions can take various forms, and each can have different legal impacts. Scholars and industry have been proposing tools to better automate the process of privacy decisions at different levels. We provide an overview of the main challenges raised by the automation of Privacy decisions....
:sunflower: Comments: 9 pages, 2 figures, accepted for publication at iwpe.info 2023
:seedling: Read more →
Keyword: edge computing (2 papers)
MetaMorphosis: Task-oriented Privacy Cognizant Feature Generation for Multi-task Learning
:brain: Authors: Md Adnan Arefeen, Zhouyu Li, Md Yusuf Sarwar Uddin, Anupam Das:paw_prints: Details: arXiv:2305.07815
:ramen: tl;dr: Running separate single-task models on edge devices is inefficient regarding the required computational resource and time. This paper proposes a novel deep learning-based privacy-cognizant feature generation process called MetaMorphosis. It outperforms recent adversarial learning and universal feature generation methods by guaranteeing privacy requirements....
:sunflower: Comments: Preprint version, 22 pages. Keywords: Multi-task learning, neural networks, collaborative intelligence, differential privacy, task privacy
:seedling: Read more →
Attention-based QoE-aware Digital Twin Empowered Edge Computing for Immersive Virtual Reality
:brain: Authors: Jiadong Yu, Ahmad Alhilal, Tailin Zhou, Hui Pan, Danny H.K. Tsang:paw_prints: Details: arXiv:2305.08569
:ramen: tl;dr: Metaverse applications such as virtual reality (VR) content streaming, require optimal resource allocation strategies for mobile edge computing (MEC) In contrast to online reinforcement learning (RL) algorithms, which can incur substantial communication overheads and longer delays, the majority of existing works employ offline-trained RL algorithms for resource allocation decisions....
:seedling: Read more →
Keyword: Blockchain (1 papers)
Interchain Timestamping for Mesh Security
:brain: Authors: Ertem Nusret Tas, Runchao Han, David Tse, Fisher Yu, Kamilla Nazirkhanova:paw_prints: Details: arXiv:2305.07830
:ramen: tl;dr: Interchain timestamping can be readily implemented for Cosmos chains connected by the Inter-Blockchain Communication (IBC) protocol. We show that it extracts the maximum economic security from the provider chains, as quantified by the slashable safety resilience. We compare interchain timestamps with cross-staking, the original solution to mesh security, and with Trustboost....
:sunflower: Comments: Forthcoming in ACM Conference on Computer and Communications Security (CCS) 2023
:seedling: Read more →
Keyword: Federated Learning (8 papers)
Understanding Model Averaging in Federated Learning on Heterogeneous Data
:brain: Authors: Tailin Zhou, Zehong Lin, Jun Zhang, Danny H.K. Tsang:paw_prints: Details: arXiv:2305.07845
:ramen: tl;dr: Model averaging is a widely adopted technique in federated learning (FL) It aggregates multiple client models trained on heterogeneous data to obtain a well-performed global model. We decompose the expected prediction error of the global model into five factors related to client models. We propose adopting iterative moving averaging (IMA) on global models....
:seedling: Read more →
A Federated Learning-based Industrial Health Prognostics for Heterogeneous Edge Devices using Matched Feature Extraction
:brain: Authors: Anushiya Arunan, Yan Qin, Xiaoli Li, Chau Yuen:paw_prints: Details: arXiv:2305.07854
:ramen: tl;dr: Industrial health prognostics require rich training data to develop accurate and reliable predictive models. The abundance of edge industrial data necessitate decentralized data utilization. We propose a pioneering FL-based health prognostic model with a feature similarity-matched parameter aggregation algorithm to discriminately learn from heterogeneous edge data....
:sunflower: Comments: This paper has been accepted by IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING
:seedling: Read more →
A Survey of Federated Evaluation in Federated Learning
:brain: Authors: Behnaz Soltani, Yipeng Zhou, Venus Haghighi, John C.S. Lui:paw_prints: Details: arXiv:2305.08070
:ramen: tl;dr: In traditional machine learning, it is trivial to conduct model evaluation since all data samples are managed centrally by a server. In federated learning (FL), clients do not expose their original data to preserve data privacy. Federated evaluation plays a vital role in client selection, incentive mechanism design, malicious attack detection, etc....
:seedling: Read more →
Federated TD Learning over Finite-Rate Erasure Channels: Linear Speedup under Markovian Sampling
:brain: Authors: Nicolò Dal Fabbro, Aritra Mitra, George J. Pappas:paw_prints: Details: arXiv:2305.08104
:ramen: tl;dr: Federated learning (FL) has recently gained much attention due to its effectiveness in speeding up supervised learning tasks. However, whether similar speedups can be established for reinforcement learning remains much less understood theoretically. We propose and analyze QFedTD - a quantized federated temporal difference learning algorithm with linear function approximation....
:seedling: Read more →
Balancing Privacy and Utility of Spatio-Temporal Data for Taxi-Demand Prediction
:brain: Authors: Yumeki Goto, Tomoya Matsumoto, Hamada Rizk, Naoto Yanai, Hirozumi Yamaguchi:paw_prints: Details: arXiv:2305.08107
:ramen: tl;dr: Taxi-demand prediction is an important application of machine learning. The use of sensitive data in these systems raises concerns about privacy and security. We propose a system that utilizes region-independent encoding for geographic lat-long coordinates. We employ cost-sensitive learning and various regularization techniques to mitigate issues related to data scarcity and overfitting....
:seedling: Read more →
FedAds: A Benchmark for Privacy-Preserving CVR Estimation with Vertical Federated Learning
:brain: Authors: Penghui Wei, Hongjian Dou, Shaoguo Liu, Rongjun Tang, Li Liu, Liang Wang, Bo Zheng:paw_prints: Details: arXiv:2305.08328
:ramen: tl;dr: Conversion rate (CVR) estimation aims to predict the probability of conversion event after a user has clicked an ad. To estimate CVR accurately and protect data privacy better, vertical federated learning (vFL) is a natural solution. FedAds is the first benchmark for CVR estimation with vFL....
:sunflower: Comments: SIGIR 2023, Resource Track
:seedling: Read more →
Quadratic Functional Encryption for Secure Training in Vertical Federated Learning
:brain: Authors: Shuangyi Chen, Anuja Modi, Shweta Agrawal, Ashish Khisti:paw_prints: Details: arXiv:2305.08358
:ramen: tl;dr: Vertical federated learning (VFL) enables the collaborative training of machine learning (ML) models. In VFL, the labels are available to a single party and the complete feature set is formed only when data from all parties is combined. Xu et al. proposed a new framework called FedV for secure gradient computation for VFL....
:sunflower: Comments: Accepted by ISIT 2023
:seedling: Read more →
ESAFL: Efficient Secure Additively Homomorphic Encryption for Cross-Silo Federated Learning
:brain: Authors: Jiahui Wu, Weizhe Zhang, Fucai Luo:paw_prints: Details: arXiv:2305.08599
:ramen: tl;dr: Cross-silo federated learning (FL) enables multiple clients to collaboratively train a machine learning model without sharing training data. Techniques using homomorphic encryption (HE) have been designed to solve this but bring their own challenges. We present a new FL protocol ESAFL that leverages a novel efficient and secure additively HE....
:seedling: Read more →