Decoupled Classifier-Free Guidance for Counterfactual Diffusion Models
This addresses a key limitation in counterfactual generation for causal inference applications, offering a model-agnostic solution to improve precision, though it is incremental as it builds on existing diffusion model techniques.
The paper tackles the problem of spurious changes in counterfactual generation using diffusion models by identifying that classifier-free guidance applies a global scale to all attributes, and proposes Decoupled Classifier-Free Guidance (DCFG) to enable attribute-wise control, resulting in more accurate and disentangled counterfactuals.
Counterfactual generation aims to simulate realistic hypothetical outcomes under causal interventions. Diffusion models have emerged as a powerful tool for this task, combining DDIM inversion with conditional generation and classifier-free guidance (CFG). In this work, we identify a key limitation of CFG for counterfactual generation: it prescribes a global guidance scale for all attributes, leading to significant spurious changes in inferred counterfactuals. To mitigate this, we propose Decoupled Classifier-Free Guidance (DCFG), a flexible and model-agnostic guidance technique that enables attribute-wise control following a causal graph. DCFG is implemented via a simple attribute-split embedding strategy that disentangles semantic inputs, enabling selective guidance on user-defined attribute groups.