Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Filters








36 Hits in 5.3 sec

Security-Preserving Federated Learning via Byzantine-Sensitive Triplet Distance [article]

Youngjoon Lee, Sangwoo Park, Joonhyuk Kang
2022 arXiv   pre-print
While being an effective framework of learning a shared model across multiple edge devices, federated learning (FL) is generally vulnerable to Byzantine attacks from adversarial edge devices.  ...  Key idea is to extract essential information from every local models along with the previous global model to define a distance measure in a manner similar to triplet loss.  ...  Based on the proposed triplet distance that uses the previous round's global model as an anchor, the proposed FL is robust to Byzantine attacks. model poisoning attacks.  ... 
arXiv:2210.16519v1 fatcat:a5o3p4z5obdhdes2pox6vzepue

Privacy and Robustness in Federated Learning: Attacks and Defenses [article]

Lingjuan Lyu, Han Yu, Xingjun Ma, Chen Chen, Lichao Sun, Jun Zhao, Qiang Yang, Philip S. Yu
2022 arXiv   pre-print
Finally, we discuss promising future research directions towards robust and privacy-preserving federated learning.  ...  Recently, federated learning (FL) has emerged as an alternative solution and continue to thrive in this new reality.  ...  [30] , [40] , [88] Secure Multiparty Computation [89] , [5] While privacy preservation has been extensively studied in the machine learning community, privacy preservation in federated learning  ... 
arXiv:2012.06337v3 fatcat:f5aflxnsdrdcdf4kvoa6yzseqq

Federated Unlearning: A Survey on Methods, Design Guidelines, and Evaluation Metrics [article]

Nicolò Romandini, Alessio Mora, Carlo Mazzocca, Rebecca Montanari, Paolo Bellavista
2024 arXiv   pre-print
Federated Learning (FL) enables collaborative training of a Machine Learning (ML) model across multiple parties, facilitating the preservation of users' and institutions' privacy by keeping data stored  ...  This highlights the necessity for novel Federated Unlearning (FU) algorithms, which can efficiently remove specific clients' contributions without full model retraining.  ...  to preserve privacy and/or enhance security.  ... 
arXiv:2401.05146v2 fatcat:u4pvk3obkvfipo2lnkq4hvskpm

D1.1 - State of the Art Analysis

Danilo Ardagna
2021 Zenodo  
Then, the deliverable provides a background on AI applications design, also considering some advanced design trends (e.g., Network Architecture Search, Federated Learning, Deep Neural Networks partitioning  ...  The aim of the AI-SPRINT "Artificial intelligence in Secure PRIvacy-preserving computing coNTinuum" project is to develop a platform composed of design and runtime management tools to seamlessly design  ...  As part of the contribution to federated learning, AI-SPRINT will design methods for the seamless update of models either via weights updates or via model redeployment.  ... 
doi:10.5281/zenodo.6372377 fatcat:f6ldfuwivbcltew4smiiwphfty

Converging blockchain and next-generation artificial intelligence technologies to decentralize and accelerate biomedical research and healthcare

Polina Mamoshina, Lucy Ojomoko, Yury Yanovich, Alex Ostrovski, Alex Botezatu, Pavel Prikhodko, Eugene Izumchenko, Alexander Aliper, Konstantin Romantsov, Alexander Zhebrak, Iraneus Obioma Ogu, Alex Zhavoronkov
2017 OncoTarget  
A secure and transparent distributed personal data marketplace utilizing blockchain and deep learning technologies may be able to resolve the challenges faced by the regulators and return the control over  ...  The novel deep learning and transfer learning techniques are turning any data about the person into medical data transforming simple facial pictures and videos into powerful sources of data for predictive  ...  The choice of constant K depends on the blockchain security model; as per Byzantine fault tolerance assumptions, K > round (N/3). 2.  ... 
doi:10.18632/oncotarget.22345 pmid:29464026 pmcid:PMC5814166 fatcat:letjwuibxfd5tosha5qipuyiky

Hidden in Plain Sight: Storing and Managing Secrets on a Public Ledger [article]

Eleftherios Kokoris-Kogias, Enis Ceyhun Alp, Sandra Deepthy Siby, Nicolas Gailly, Philipp Jovanovic, Linus Gasser, Bryan Ford
2018 IACR Cryptology ePrint Archive  
Distributed ledgers provide high availability and integrity, making them a key enabler for practical and secure computation of distributed workloads among mutually distrustful parties.  ...  Our benchmarks show that transaction-processing latency increases linearly in terms of security (number of trustees) and is in the range of 0.2 to 8 seconds for 16 to 128 trustees.  ...  are, especially in a Byzantine environment.  ... 
dblp:journals/iacr/Kokoris-KogiasA18 fatcat:55uesk23ljd35a4fe3auonilka

Table of contents

2021 ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)  
AUTHENTICATION IN CLOUD Mengdi Wang, Di Xiao, Jia Liang, Chongqing University, PRIVACY-PRESERVING NEAR NEIGHBOR SEARCH VIA SPARSE CODING ................................... 2635 WITH AMBIGUATION Behrooz  ...  LINK PREDICTION VIA REINFORCEMENT LEARNING ......................................... 3460 Ye Tao, Ying Li, Zhonghai Wu, Peking University, China MLSP-26.4: A DEEP REINFORCEMENT LEARNING APPROACH TO AUDIO-BASED  ... 
doi:10.1109/icassp39728.2021.9414617 fatcat:m5ugnnuk7nacbd6jr6gv2lsfby

Applications of Economic and Pricing Models for Wireless Network Security: A Survey [article]

Nguyen Cong Luong, Dinh Thai Hoang, Ping Wang, Dusit Niyato, Zhu Han
2017 arXiv   pre-print
This paper provides a comprehensive literature review on applications of economic and pricing theory to security issues in wireless networks.  ...  Finally, we highlight important challenges, open issues and future research directions of applying economic and pricing models to wireless security issues.  ...  The bidders then submit their vectors to the auctioneer via a secure channel employing the auctioneer's public key.  ... 
arXiv:1707.07846v1 fatcat:xb3jvolwmjad7oxdefdd2pq7dq

Tutorial-Cum-Survey on Semantic and Goal- Oriented Communication: Research Landscape, Challenges, and Future Directions [article]

Tilahun M. Getu, Georges Kaddoum, Mehdi Bennis
2023 arXiv   pre-print
In mitigating this challenge in part, causal representation learning is useful, as discussed below. 5) Causal Representation Learning: semantic representation via causal representation learning [146]  ...  NeuroComm's end-to-end design [307] is based on supervised learning via surrogate gradient descent (GD) methods.  ... 
arXiv:2308.01913v1 fatcat:ia6bdxdndvhxzfams4j6rhnubi

Migrating heritage: experiences of cultural networks and cultural dialogue in Europe

Philip Long
2016 International Journal of Heritage Studies (IJHS)  
On the one hand, the visitors learn something about the varied, individual characters and biographies of those interviewed via the personal tours.  ...  The programme was intended to open up museological practice and mainstream Byzantine and post-Byzantine heritage to the Roma now living in Greece.  ...  Finally, I would like to acknowledge the insights received from the scholars and creative practitioners of the Ariadne Project, a Life Long Learning initiative researching the role of art for intercultural  ... 
doi:10.1080/13527258.2016.1182052 fatcat:rwb5jcohrjfo5delouext3bajm

Migrating heritage: Experiences of cultural networks and cultural dialogue in Europe

Gillian Oliver
2015 The Australian Library Journal  
On the one hand, the visitors learn something about the varied, individual characters and biographies of those interviewed via the personal tours.  ...  The programme was intended to open up museological practice and mainstream Byzantine and post-Byzantine heritage to the Roma now living in Greece.  ...  Finally, I would like to acknowledge the insights received from the scholars and creative practitioners of the Ariadne Project, a Life Long Learning initiative researching the role of art for intercultural  ... 
doi:10.1080/00049670.2015.1040359 fatcat:7cff7ntlnvhslm4vscymo4xcfy

Ofelimos: Combinatorial Optimization via Proof-of-Useful-Work \\ A Provably Secure Blockchain Protocol [article]

Matthias Fitzi, Aggelos Kiayias, Giorgos Panagiotakos, Alexander Russell
2021 IACR Cryptology ePrint Archive  
We provide a thorough security analysis of our protocol and additionally present metrics that reect the usefulness of the system.  ...  primitive in Nakamoto's longest chain protocol with a proof of useful work (PoUW) has been long theorized as an ideal solution in many respects but, to this day, the concept still lacks a convincingly secure  ...  Note that deviating from this strategy may only impact usefulnessthe security of the protocol is maintained against any Byzantine deviation.  ... 
dblp:journals/iacr/FitziKPR21 fatcat:vgccmerafjg5jhfdn4ekcahhqa

Abstracts of Presentations

1995 Lake and reservoir management  
the most poorly on the learning task.  ...  From the Byzantine to the Gothic, from Istria to the Levant, the journey of this stone is the journey of Venice itself.  ... 
doi:10.1080/07438149509354371 fatcat:7ugmrjclcreqflltganfbwzene

Abstracts of Presentations

1990 Journal of Aerosol Medicine  
the most poorly on the learning task.  ...  From the Byzantine to the Gothic, from Istria to the Levant, the journey of this stone is the journey of Venice itself.  ... 
doi:10.1089/jam.1990.3.suppl_1.s-3 fatcat:dknzzdwe5vecnfuinlqqvepnxe

Abstracts of Presentations

1989 Biomaterials Artificial Cells and Artificial Organs  
the most poorly on the learning task.  ...  From the Byzantine to the Gothic, from Istria to the Levant, the journey of this stone is the journey of Venice itself.  ... 
doi:10.3109/10731198909117643 fatcat:zf2t5ocbkrb2fof4k34gbqfzu4
« Previous Showing results 1 — 15 out of 36 results