TASER: Task-Aware Spectral Energy Refine for Backdoor Suppression in UAV Swarms Decentralized Federated Learning

Researchers propose TASER, a new defense framework against backdoor attacks in UAV-based decentralized federated learning systems. The system uses spectral energy analysis rather than traditional outlier detection, achieving below 20% attack success rates while maintaining accuracy within 5% loss.

• →TASER introduces the first efficient backdoor defense using spectral concentration instead of complex outlier detection methods.
• →The framework specifically targets stealthy backdoor attacks in UAV swarm federated learning environments where resources are limited.
• →Research reveals that more sophisticated backdoor attacks create more distinct spectral concentration patterns, making them detectable.
• →The system achieves strong defensive performance with attack success rates below 20% and accuracy loss under 5%.
• →TASER preserves main-task-relevant frequency coefficients while discarding others to structurally disrupt backdoor tasks.