mlsanityData to 7 Oct 2026

Mitigating attribute amplification in counterfactual image generation

arXiv 2403.09422 ↗First posted 14 Mar 2024cs.CV cs.AIcited by 12

Not on the map. Why?

Abstract (arXiv)

Causal generative modelling is gaining interest in medical imaging due to its ability to answer interventional and counterfactual queries. Most work focuses on generating counterfactual images that look plausible, using auxiliary classifiers to enforce effectiveness of simulated interventions. We investigate pitfalls in this approach, discovering the issue of attribute amplification, where unrelated attributes are spuriously affected during interventions, leading to biases across protected characteristics and disease status. We show that attribute amplification is caused by the use of hard labels in the counterfactual training process and propose soft counterfactual fine-tuning to mitigate this issue. Our method substantially reduces the amplification effect while maintaining effectiveness of generated images, demonstrated on a large chest X-ray dataset. Our work makes an important advancement towards more faithful and unbiased causal modelling in medical imaging.

Key facts. Cited by 12 papers in the map (data through 7 Oct 2026). Builds on 21 of its 29 references that are in the map. First posted on arXiv 14 Mar 2024.

Builds on

21 of 29 references.

References · 21

  1. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  2. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 8,964
  3. 2021Diffusion Models Beat GANs on Image Synthesiscited by 8,185
  4. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  5. 2022Generative Adversarial Networkscited by 5,603
  6. 2015Variational Inference with Normalizing Flowscited by 2,161
  7. 2014Conditional Generative Adversarial Netscited by 2,152
  8. 2017Causal Effect Inference with Deep Latent-Variable Modelscited by 408
  9. 2020Deep Structural Causal Models for Tractable Counterfactual Inferencecited by 210
  10. 2017CausalGAN: Learning Causal Implicit Generative Models with Adversarial Trainingcited by 152
  11. 2019Learning Neural Causal Models from Unknown Interventionscited by 113
  12. 2021The Causal-Neural Connection: Expressiveness, Learnability, and Inferencecited by 97
9 more
  1. 2022Deep End-to-end Causal Inferencecited by 93
  2. 2022Diffusion Models for Causal Discovery via Topological Orderingcited by 64
  3. 2018Conditional Density Estimation with Bayesian Normalising Flowscited by 54
  4. 2022Neural Causal Models for Counterfactual Identification and Estimationcited by 54
  5. 2023High Fidelity Image Counterfactuals with Probabilistic Causal Modelscited by 53
  6. 2021Relating Graph Neural Networks to Structural Causal Modelscited by 39
  7. 2023Measuring axiomatic soundness of counterfactual image modelscited by 34
  8. 2023Towards Causal Foundation Model: on Duality between Causal Inference and Attentioncited by 17
  9. 2017Implicit Causal Models for Genome-wide Association Studiescited by 10

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 1; 2025: 5; 2026: 6.

12 papers.

  1. 2024Robust image representations with counterfactual contrastive learningcited by 12
  2. 2025Diffusion Counterfactual Generation with Semantic Abductioncited by 8
  3. 2025Towards Desiderata-Driven Design of Visual Counterfactual Explainerscited by 7
  4. 2025Counterfactual Identifiability via Dynamic Optimal Transportcited by 5
  5. 2025Segmentor-Guided Counterfactual Fine-Tuning for Locally Coherent and Targeted Image Synthesiscited by 2
  6. 2026Scaling Generative Foundation Models for Chest Radiography with Rectified Flow Transformerscited by 2
  7. 2026An Interpretable Local Editing Model for Counterfactual Medical Image Generationcited by 1
  8. 2026When the Edit Changes the Patient: Measuring Identity Preservation in Counterfactual Retinal Imagescited by 1
  9. 2025Diffusion-based Counterfactual Augmentation: Towards Robust and Interpretable Knee Osteoarthritis Gradingcited by 0
  10. 2026Positional Segmentor-Guided Counterfactual Fine-Tuning for Spatially Localized Image Synthesiscited by 0
  11. 2026Generating Medical Image Counterfactuals using Causal Explanationscited by 0
  12. 2026A Data-Interventional Framework for Auditing Privacy and Fairness in Generative Medical Imagingcited by 0