Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
The paper introduces Opportunistic Target Selection (OTS), a wrapper for black-box adversarial attacks that dynamically switches to a targeted objective based on the most likely non-true class during the attack's early stages. This mitigates class drift, improving query efficiency by focusing the search towards a specific adversarial goal. Experiments on ImageNet classifiers show that OTS significantly boosts the success rate and reduces the number of queries for random-search attacks like SimBA and Square Attack, particularly on standard ResNet-50 models.
Black-box attacks can be made dramatically more efficient by simply steering them towards the most promising adversarial class early on, achieving up to a 27% improvement in success rate.
Black-box adversarial attacks that minimize only the ground-truth confidence suffer from class drift: perturbations wander through the feature space without committing to a specific adversarial class, wasting queries on diffuse, undirected progress. We introduce Opportunistic Target Selection (OTS), a lightweight wrapper that switches an untargeted attack to a targeted objective early in its trajectory, locking onto whichever non-true class currently leads. OTS requires no architectural modification to the underlying attack, no gradient access, and no a priori target-class knowledge. We validate OTS on three score-based attacks (SimBA, Square Attack with cross-entropy loss, and Bandits) across five standard ImageNet classifiers (4,500 runs). On random-search attacks, OTS closely tracks oracle performance, with gains up to +27 pp in success rate and 43% relative reduction in censored-mean iterations on ResNet-50. On gradient-estimation attacks (Bandits) and attacks with margin loss, OTS is redundant, a negative result that reinforces our interpretation of OTS as a margin-loss surrogate. On adversarially-trained models, a bimodal difficulty distribution eliminates the regime where targeting helps.