Dim-Krum: Backdoor-Resistant Federated Learning for NLP with Dimension-wise Krum-Based Aggregation

10/13/2022
by   Zhiyuan Zhang, et al.
5

Despite the potential of federated learning, it is known to be vulnerable to backdoor attacks. Many robust federated aggregation methods are proposed to reduce the potential backdoor risk. However, they are mainly validated in the CV field. In this paper, we find that NLP backdoors are hard to defend against than CV, and we provide a theoretical analysis that the malicious update detection error probabilities are determined by the relative backdoor strengths. NLP attacks tend to have small relative backdoor strengths, which may result in the failure of robust federated aggregation methods for NLP attacks. Inspired by the theoretical results, we can choose some dimensions with higher backdoor strengths to settle this issue. We propose a novel federated aggregation algorithm, Dim-Krum, for NLP tasks, and experimental results validate its effectiveness.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
12/24/2019

Attack-Resistant Federated Learning with Residual-based Reweighting

Federated learning has a variety of applications in multiple domains by ...
research
07/18/2021

RobustFed: A Truth Inference Approach for Robust Federated Learning

Federated learning is a prominent framework that enables clients (e.g., ...
research
03/20/2023

Recursive Euclidean Distance Based Robust Aggregation Technique For Federated Learning

Federated learning has gained popularity as a solution to data availabil...
research
01/31/2022

Securing Federated Sensitive Topic Classification against Poisoning Attacks

We present a Federated Learning (FL) based solution for building a distr...
research
06/21/2020

Free-rider Attacks on Model Aggregation in Federated Learning

Free-rider attacks on federated learning consist in dissimulating partic...
research
01/15/2021

Probabilistic Inference for Learning from Untrusted Sources

Federated learning brings potential benefits of faster learning, better ...
research
10/08/2022

Collaborative Domain Blocking: Using federated NLP To Detect Malicious Domains

Current content filtering and blocking methods are susceptible to variou...

Please sign up or login with your details

Forgot password? Click here to reset