# Beyond Corner Patches: Semantics-Aware Backdoor Attack in Federated Learning Source: [Beyond Corner Patches: Semantics-Aware Backdoor Attack in Federated Learning]() Collection index: [llms.txt]() Data updated: 2026-04-09T05:00:04.534Z Authors: Kavindu Herath, Joshua Zhao, Saurabh Bagchi Published: 2026-03-31T06:48:25.000Z ## Abstract Backdoor attacks on federated learning (FL) are most often evaluated with synthetic corner patches or out-of-distribution (OOD) patterns that are unlikely to arise in practice. In this paper, we revisit the backdoor threat to standard FL (a single global model) under a more realistic setting where triggers must be semantically meaningful, in-distribution, and visually plausible. We propose SABLE, a Semantics-Aware Backdoor for LEarning in federated settings, which constructs natural, content-consistent triggers (e.g., semantic attribute changes such as sunglasses) and optimizes an aggregation-aware malicious objective with feature separation and parameter regularization to keep attacker updates close to benign ones. We instantiate SABLE on CelebA hair-color classification and the German Traffic Sign Recognition Benchmark (GTSRB), poisoning only a small, interpretable subset of each malicious client's local data while otherwise following the standard FL protocol. Across heterogeneous client partitions and multiple aggregation rules (FedAvg, Trimmed Mean, MultiKrum, and FLAME), our semantics-driven triggers achieve high targeted attack success rates while preserving benign test accuracy. These results show that semantics-aligned backdoors remain a potent and practical threat in federated learning, and that robustness claims based solely on synthetic patch triggers can be overly optimistic. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Traffic Sign Recognition on GTSRB](): Traffic Sign Recognition; GTSRB; Accuracy: 100. - [Hair Color Classification on CelebA](): Hair color classification; CelebA; Accuracy: 90.8. ## Original sources and implementation [Original paper]() [PDF]()