Retrievalguard: Provably Robust 1-nearest Neighbor Image Retrieval

Abstract

Recent research works have shown that image retrieval models are vulnerable to adversarial attacks, where slightly modified test inputs could lead to problematic retrieval results. In this paper, we aim to design a provably robust image retrieval model which keeps the most important evaluation metric Recall@1 invariant to adversarial perturbation. We propose the first 1-nearest neighbor (NN) image retrieval algorithm, RetrievalGuard, which is provably robust against adversarial perturbations within an \(ℓ₂\) ball of calculable radius. The challenge is to design a provably robust algorithm that takes into consideration the 1-NN search and the high-dimensional nature of the embedding space. Algorithmically, given a base retrieval model and a query sample, we build a smoothed retrieval model by carefully analyzing the 1-NN search procedure in the high-dimensional embedding space. We show that the smoothed retrieval model has bounded Lipschitz constant and thus the retrieval score is invar

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keywu2022retrievalguard

Related papers

• Defense Against Adversarial Images Using Web-scale Nearest-neighbor Search (2019)10.97
• Risk Controlled Image Retrieval (2023)0.00
• An Alternative Proof Of The Vulnerability Of Retrieval In High Intrinsic Dimensionality Neighborhood (2020)0.00
• Boundary-aware Backward-compatible Representation Via Adversarial Learning In Image Retrieval (2023)9.21
• Leveraging Reinforcement Learning For Evaluating Robustness Of KNN Search Algorithms (2021)0.00
• Adversarial Attack On Deep Product Quantization Network For Image Retrieval (2020)11.29
• Who's Afraid Of Adversarial Queries? The Impact Of Image Modifications On Content-based Image Retrieval (2019)0.00
• Query Attack Via Opposite-direction Feature:towards Robust Image Retrieval (2018)0.00