Search papers, labs, and topics across Lattice.
This paper introduces BMAT (Bilevel-Minimax Adversarial Transfer), a novel approach to transfer-based adversarial attacks that utilizes a bilevel-minimax optimization framework to enhance the transferability of adversarial examples across different model architectures. By addressing the interdependencies between initialization, surrogate model adaptation, and perturbation optimization, BMAT significantly improves robustness and generalization in adversarial settings. Experimental results demonstrate that BMAT outperforms over 10 state-of-the-art baselines, achieving up to a 2x reduction in mean Intersection over Union (mIoU) across more than 30 victim models in classification and segmentation tasks.
BMAT achieves unprecedented transferability in adversarial attacks, outperforming over 10 strong baselines and cutting mIoU losses in half.
Transfer-based adversarial attacks craft adversarial examples using surrogate models to mislead black-box victim models. Beyond perturbation generation, transferability is fundamentally governed by the coupling of initialization, surrogate adaptation, and gradient dynamics. We revisit this challenge from a bilevel-minimax perspective and propose BMAT (Bilevel-Minimax Adversarial Transfer). The bilevel formulation captures the dependency between initialization and perturbation, while the inner minimax problem promotes surrogate robustness for cross-architecture generalization. Algorithmically, we develop an integrated bottom-up solver that combines a Soft Weight Modulator and an Implicit Gradient Approximator to enable ternary coupling among initialization, surrogate adaptation, and perturbation optimization. We further provide theoretical insights into the optimization dynamics of the proposed bilevel-minimax framework. Extensive experiments on classification and segmentation benchmarks show that BMAT outperforms more than 10 strong baselines across more than 30 victim models, improving both intra- and cross-architecture transfer and yielding up to a 2x reduction in mIoU. Code is available at https://github.com/callous-youth/BMAT.
