Virat Shejwalkar (UMass Amherst), Amir Houmansadr (UMass Amherst)

Federated learning (FL) enables many data owners (e.g., mobile devices) to train a joint ML model (e.g., a next-word prediction classifier) without the need of sharing their private training data.

However, FL is known to be susceptible to poisoning attacks by malicious participants (e.g., adversary-owned mobile devices) who aim at hampering the accuracy of the jointly trained model through sending malicious inputs during the federated training process.

In this paper, we present a generic framework for model poisoning attacks on FL. We show that our framework leads to poisoning attacks that substantially outperform state-of-the-art model poisoning attacks by large margins. For instance, our attacks result in $1.5times$ to $60times$ higher reductions in the accuracy of FL models compared to previously discovered poisoning attacks.

Our work demonstrates that existing Byzantine-robust FL algorithms are significantly more susceptible to model poisoning than previously thought. Motivated by this, we design a defense against FL poisoning, called emph{divide-and-conquer} (DnC). We demonstrate that DnC outperforms all existing Byzantine-robust FL algorithms in defeating model poisoning attacks,
specifically, it is $2.5times$ to $12times$ more resilient in our experiments with different datasets and models.

View More Papers

Reining in the Web's Inconsistencies with Site Policy

Stefano Calzavara (Università Ca' Foscari Venezia), Tobias Urban (Institute for Internet Security and Ruhr University Bochum), Dennis Tatang (Ruhr University Bochum), Marius Steffens (CISPA Helmholtz Center for Information Security), Ben Stock (CISPA Helmholtz Center for Information Security)

Read More

As Strong As Its Weakest Link: How to Break...

Kai Li (Syracuse University), Jiaqi Chen (Syracuse University), Xianghong Liu (Syracuse University), Yuzhe Tang (Syracuse University), XiaoFeng Wang (Indiana University Bloomington), Xiapu Luo (Hong Kong Polytechnic University)

Read More

Oblivious DNS over HTTPS (ODoH): A Practical Privacy Enhancement...

Sudheesh Singanamalla*†, Suphanat Chunhapanya*, Jonathan Hoyland*, Marek Vavruša*, Tanya Verma*, Peter Wu*, Marwan Fayed*, Kurtis Heimerl†, Nick Sullivan*, Christopher Wood* (*Cloudflare Inc. †University of Washington)

Read More

Deceptive Deletions for Protecting Withdrawn Posts on Social Media...

Mohsen Minaei (Visa Research), S Chandra Mouli (Purdue University), Mainack Mondal (IIT Kharagpur), Bruno Ribeiro (Purdue University), Aniket Kate (Purdue University)

Read More