RADAR: Defending RAG Dynamically against Retrieval Corruption
For practitioners deploying RAG in dynamic environments, RADAR provides a robust defense against adversarial attacks without prohibitive storage costs.
RADAR defends RAG systems against retrieval corruption in dynamic settings by modeling context selection as a graph-based energy minimization problem solved via Max-Flow Min-Cut, using a Bayesian memory node to update belief states. It achieves superior robustness and response quality with minimal storage overhead compared to baselines.
While RAG systems are increasingly deployed in dynamic web search, temporal volatility amplifies their vulnerability to adversarial attacks. Existing static-oriented defenses struggle to handle evolving threats and incur prohibitive storage costs in dynamic settings. We propose RADAR, a framework that models reliable context selection as a graph-based energy minimization problem, solved exactly via Max-Flow Min-Cut. By incorporating a Bayesian memory node, RADAR recursively updates a belief state instead of archiving raw historical documents, effectively balancing stability against attacks with adaptability to genuine knowledge shifts. Experiments on a novel dynamic dataset show that RADAR achieves superior robustness and response quality with minimal storage overhead compared to the baselines.