Learning Counterfactually Decoupled Attention for Open-World Model Attribution

TL;DR

This paper introduces CDAL, a causal learning method that improves open-world model attribution by decoupling confounding factors, leading to better generalization to unseen attacks with minimal computational cost.

Contribution

We propose a novel causal decoupling approach for model attribution that enhances robustness and generalization in open-world scenarios.

Findings

Significant improvement over state-of-the-art methods on open-world benchmarks.

Effective in detecting novel and unseen attacks.

Minimal additional computational overhead.

Abstract

In this paper, we propose a Counterfactually Decoupled Attention Learning (CDAL) method for open-world model attribution. Existing methods rely on handcrafted design of region partitioning or feature space, which could be confounded by the spurious statistical correlations and struggle with novel attacks in open-world scenarios. To address this, CDAL explicitly models the causal relationships between the attentional visual traces and source model attribution, and counterfactually decouples the discriminative model-specific artifacts from confounding source biases for comparison. In this way, the resulting causal effect provides a quantification on the quality of learned attention maps, thus encouraging the network to capture essential generation patterns that generalize to unseen source models by maximizing the effect. Extensive experiments on existing open-world model attribution benchmarks show that with minimal computational overhead, our method consistently improves state-of-the-art models by large margins, particularly for unseen novel attacks. Source code: https://github.com/yzheng97/CDAL.