Research map / Image generation / Autoregressive and latent generation
STEEX: Steering Counterfactual Explanations with Semantics
Paul Jacob, Éloi Zablocki, Hédi Ben-Younes, Mickaël Chen, Patrick Pérez, Matthieu Cord
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add the distance loss L_dist (lambda=0.3 instead of 0) (ablation) | STEEX · CelebAMask-HQ Smile | FVA (%): 65.2 -> 97.6 | Improved |
| add the distance loss L_dist (lambda=0.3 instead of 0) (ablation) | STEEX · CelebAMask-HQ Young | FVA (%): 37.0 -> 96.0 | Improved |
| add the distance loss L_dist (ablation) | STEEX · CelebAMask-HQ Smile | FID: 29.7 -> 21.9 | Improved |
| add the distance loss L_dist (ablation) | STEEX · CelebAMask-HQ Young | FID: 45.7 -> 26.8 | Improved |
| use ground-truth instead of predicted segmentation masks (ablation) | STEEX · CelebAMask-HQ Smile | FID: 21.9 -> 21.2 | No clear effect |
| use ground-truth instead of predicted segmentation masks (ablation) | STEEX · CelebAMask-HQ Young | FVA (%): 96.0 -> 98.2 | No clear effect |
| replace PE/DiVE with STEEX semantic-guided generation | STEEX · CelebAMask-HQ Smile | FID: 52.4 -> 21.9 (vs PE), 107.0 -> 21.9 (vs DiVE) | Improved |
| replace PE/DiVE with STEEX semantic-guided generation | STEEX · CelebAMask-HQ Young | FVA (%): 76.2 -> 96.0 (vs PE), 32.3 -> 96.0 (vs DiVE) | Improved |
| replace PE/DiVE with STEEX semantic-guided generation | STEEX · CelebAMask-HQ Young | MNAC: 8.51 -> 5.63 (vs PE), 6.76 -> 5.63 (vs DiVE) | Improved |
| replace DiVE with STEEX on low-res CelebA | STEEX · CelebA Smile | FVA (%): 97.3 (DiVE) vs 96.9 (STEEX) | No clear effect |
| scale resolution from CelebA 128 to CelebAMask-HQ 256 | DiVE · CelebA Smile -> CelebAMask-HQ Smile | FID: 29.4 -> 107.0 | Hurt |
Closest papers by meaning
- V-CECE: Visual Counterfactual Explanations via Conceptual Edits (NeurIPS 2025 · similarity 0.89)
- Rethinking Visual Counterfactual Explanations Through Region Constraint (ICLR 2025 · similarity 0.88)
- Structure Your Data: Towards Semantic Graph Counterfactuals (ICML 2024 · similarity 0.87)
- Counterfactual Explanations for Misclassified Images: How Human and Machine Explanations Differ (Abstract Reprint) (AAAI 2024 · similarity 0.87)
- Towards Fine-Grained Interpretability: Counterfactual Explanations for Misclassification with Saliency Partition (CVPR 2025 · similarity 0.86)
- LeapFactual: Reliable Visual Counterfactual Explanation Using Conditional Flow Matching (NeurIPS 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
