Detecting Byzantine Attacks Without Clean Reference.

Note that every symbol received by the destination may be altered, and hence no clean reference observation is available to the destination. ... For this network, we identify a large family of Byzantine attacks that can be detected in the physical layer. ... It is also possible to detect Byzantine attacks without assuming any prior shared secret in certain network scenarios, where a "clean" reference is available for attack detection. ...

doi:10.1109/tifs.2016.2596140 fatcat:wmeowfwrzfes5d6h7qrsiybfva

Multiple Versions

Then we propose a new byzantine attack method called weight attack to defeat those defense schemes, and conduct experiments to demonstrate its threat. ... Finally, we indicate possible countermeasures for weight attack, and highlight several challenges and future research directions for mitigating byzantine attacks in FL. ... As a comparison, we also consider the case without attackers. ...

arXiv:2112.14468v2 fatcat:wle6xkoeqrfbtdjb6ofe7iovie

Open Access Multiple Versions

In an organized way, we then detail the existing integrity and privacy attacks as well as their defenses. ... Since the poisoned images are mislabeled, unclean label attacks can be easily detected by simple data filtering or human inspection [102] . Therefore, clean label stand-alone backdoor is proposed. ... TABLE I TAXONOMY I OF BYZANTINE AND BACKDOOR ATTACKS. ...

arXiv:2112.10183v1 fatcat:ujfz4a5mdrhsbk4kiqoqo2snfe

FLDetector aims to detect and remove the majority of the malicious clients such that a Byzantine-robust FL method can learn an accurate global model using the remaining clients. ... After removing the detected malicious clients, existing Byzantine-robust FL methods can learn accurate global models.Our code is available at https://github.com/zaixizhang/FLDetector. ... First, FLDetector addresses the limitations of existing detection methods such as the requirement of clean validation datasets. ...

arXiv:2207.09209v4 fatcat:hvjwl5msebhavim5s45tjqcsia

Open Access Multiple Versions

In this paper, we explore the threat of poisoning attacks and introduce a game-based robust federated averaging algorithm to detect and discard bad updates provided by the clients. ... Federated learning techniques are considerably vulnerable to poisoning attacks. ... Authors in [8] proposed a byzantine-robust aggregation algorithm, referred to as KRUM, which is based on the similarity of the client updates. ...

doi:10.5829/ije.2021.34.04a.09 fatcat:czvb2lifwfbkrjuiwry5mcgdna

We propose Zeno++, a new robust asynchronous Stochastic Gradient Descent~(SGD) procedure which tolerates Byzantine failures of the workers. ... We prove the convergence of Zeno++ for non-convex problems under Byzantine failures. Experimental results show that Zeno++ outperforms existing approaches. ... Baselines We use the asynchronous SGD without attacks as the gold standard, referred to as AsyncSGD without attack. ...

arXiv:1903.07020v5 fatcat:6cgsllyf75akdcgjuiza64xuny

Multiple Versions

Through a concise introduction to the concept of FL, and a unique taxonomy covering: 1) threat models; 2) poisoning attacks and defenses against robustness; 3) inference attacks and defenses against privacy ... We highlight the intuitions, key techniques as well as fundamental assumptions adopted by various attacks and defenses. ... This makes them susceptible to poisoning attacks, as the adversary can make small but damaging changes in the highdimensional models without being detected. ...

arXiv:2012.06337v3 fatcat:f5aflxnsdrdcdf4kvoa6yzseqq

Open Access Multiple Versions

However, the inaccessible local training data and uninspectable local training process make FL susceptible to various Byzantine attacks (e.g., data poisoning and model poisoning attacks), aiming to manipulate ... Most of the existing Byzantine-robust FL schemes cannot effectively defend against stealthy poisoning attacks that craft poisoned models statistically similar to benign models. ... Figure 3 : 3 Figure 2: Accuracies without Byzantine attacks. Figure 5 : 5 Figure 5: ASR versus backdoor sample percentage. ...

doi:10.1609/aaai.v36i8.20903 fatcat:4xahbrbqqnaa5dykg36k6wc2ye

Using Blades, we re-evaluate representative attacks and defenses on wide-ranging experimental configurations (approximately 1,500 trials in total). ... Despite the plethora of research focusing on Byzantine-resilient FL, the academic community has yet to establish a comprehensive benchmark suite, pivotal for impartial assessment and comparison of different ... Attacks at Level 2 and higher are referred to as "Byzantine attacks" [39] because the updates submitted by clients can be arbitrary. ...

arXiv:2206.05359v4 fatcat:72qkijaxpzf2tm4fv3bw4hkejq

Multiple Versions

In this work, we propose FLANDERS, a novel federated learning (FL) aggregation scheme robust to Byzantine attacks. ... Furthermore, FLANDERS remains highly effective even under extremely severe attack scenarios, as opposed to existing defense strategies. ... series anomaly detection. ...

arXiv:2303.16668v1 fatcat:ogkvik5oyveyzkzwgh2m5vpx44

Adversarial attacks against data privacy, learning algorithm stability, and system confidentiality are particularly concerning in the context of distributed training in federated learning. ... Hence, previously discussed robust aggregation methods and Byzantine detection schemes could defend against backdoor attacks to some extent. ... Recently, BytoChain [121] introduces a Byzantine resistant secure blockchained federated learning framework, which executes heavy verification workflows in parallel and detects byzantine attacks through ...

arXiv:2302.10637v1 fatcat:yeqkzgz6krhxnpsgeqrvitrz4u

Byzantine attacks, and irregular data distribution during training; (2) privacy protection during distributed learning and model inference at deployment; and (3) AI fairness and governance with respect ... Then we provide a unique taxonomy of countermeasures for trustworthy distributed AI, covering (1) robustness to evasion attacks and irregular queries at inference, and robustness to poisoning attacks, ... explicit detection [24] , (2) Auto-detection without auto-repair [25, 26] , and (3) Auto-detect followed by auto-repair [27] . ...

arXiv:2402.01096v1 fatcat:r6h3ciftzzcsvfu76wjmj5e3pm

Open Access

Although the implementation of these systems usually requires the use of Byzantine fault-tolerant (BFT) protocols, they are not a complete solution. ... to clean the intrusion. ... To deal with these vulnerabilities it is necessary to increment the replicated system with a synchronous subsystem capable of triggering timely recoveries without interference of attackers [13] . ...

doi:10.1109/dsnw.2011.5958857 dblp:conf/dsn/Bessani11 fatcat:q32abxyp6feghij6k7rokekqa4

Our secret sauce is a new communication abstraction called detectable all-to-all (DA2A), which allows us to detect players (byzantine or rational) that deviate from the protocol and punish them. ... A key component in permissioned blockchain protocols is a byzantine fault tolerant (BFT) consensus engine that orders transactions. ... Our protocol features the first byzantine fault-tolerant consensus engine to ensure fairness when all players are rational. It is also simple to understand and implement. ...

doi:10.4230/lipics.opodis.2019.4 dblp:conf/opodis/Lev-AriSKM19 fatcat:xuglrze6znfzhgox7qv56diaxy

FL enables efficient model generation from local data storage of the edge devices without revealing the sensitive data to any entities. ... Further, we extensively analyze the cyber threats that could be observed in a resource-constrained FL environment, and how blockchain can play a key role to block those cyber attacks. ... pattern appears, the model predicts the target label, otherwise, behaves normally[33].Backdoor attacks are also referred to as Trojan attacks.Adversaries inject clean or dirty backdoor updates into data ...

arXiv:2303.13727v1 fatcat:nwrlxlmvfjcaldvfoamyxe4tfu

Open Access

Detecting Byzantine Attacks Without Clean Reference

Preserved Fulltext

Challenges and Approaches for Mitigating Byzantine Attacks in Federated Learning [article]

Preserved Fulltext

Other Versions

Robust and Privacy-Preserving Collaborative Learning: A Comprehensive Survey [article]

Preserved Fulltext

FLDetector: Defending Federated Learning Against Model Poisoning Attacks via Detecting Malicious Clients [article]

Preserved Fulltext

Other Versions

A Game-Theoretic Approach for Robust Federated Learning

Preserved Fulltext

Zeno++: Robust Fully Asynchronous SGD [article]

Preserved Fulltext

Other Versions

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

Preserved Fulltext

Other Versions

FedInv: Byzantine-Robust Federated Learning by Inversing Local Model Updates

Preserved Fulltext

Blades: A Unified Benchmark Suite for Byzantine Attacks and Defenses in Federated Learning [article]

Preserved Fulltext

Other Versions

A Byzantine-Resilient Aggregation Scheme for Federated Learning via Matrix Autoregression on Client Updates [article]

Preserved Fulltext

A Survey of Trustworthy Federated Learning with Perspectives on Security, Robustness, and Privacy [article]

Preserved Fulltext

Trustworthy Distributed AI Systems: Robustness, Privacy, and Governance [article]

Preserved Fulltext

From Byzantine fault tolerance to intrusion tolerance (a position paper)

Preserved Fulltext

FairLedger: A Fair Blockchain Protocol for Financial Institutions

Preserved Fulltext

A Survey on Secure and Private Federated Learning Using Blockchain: Theory and Application in Resource-constrained Computing [article]

Preserved Fulltext