Learnijoy
ResearchAI Engineering & DevToolsAI Research

Counterfactual Fairness Does Not Imply Group Fairness in Images

Sangwon Jung, Sumin Yu, Sanghyuk Chun, Taesup Moon· July 9, 2026 View original

Summary

A study using newly constructed image datasets found that counterfactual fairness does not necessarily imply group fairness in image classification, contrary to observations in tabular data. Researchers propose Counterfactual Knowledge Distillation (CKD) to mitigate correlations with latent attributes, improving group fairness.

The relationship between counterfactual fairness (CF) and group fairness (GF) in algorithmic systems has been a subject of active research. While previous studies on tabular datasets suggested that CF might imply GF, this paper investigates this relationship specifically within image classification tasks, where collecting counterfactual samples for sensitive attributes is challenging. To address this, researchers created new image datasets, `oursceleb` and `ourslfw`, by employing high-quality image editing and human annotation. These datasets allow for simultaneous evaluation of both CF and GF. Empirical observations revealed that, unlike tabular data, CF does not imply GF in image classification. The study theoretically attributes this divergence to the existence of latent attributes correlated with, but not caused by, the sensitive attribute. Based on this, a simple baseline called Counterfactual Knowledge Distillation (CKD) is proposed to reduce reliance on these latent correlations. Experiments show that CF-achieving models can satisfy GF if the influence of these latent attributes is successfully reduced using methods like CKD.

Why it matters

Professionals developing or deploying image classification AI must understand that achieving one type of fairness (counterfactual) does not automatically guarantee another (group fairness), necessitating a multi-faceted approach to ethical AI. This impacts regulatory compliance and public trust.

How to implement this in your domain

  1. 1Assess existing image classification models for both counterfactual and group fairness metrics.
  2. 2Consider the potential for latent attributes to influence fairness outcomes in your datasets.
  3. 3Explore techniques like Counterfactual Knowledge Distillation (CKD) to mitigate unintended correlations.
  4. 4Develop diverse and carefully annotated datasets that allow for robust fairness evaluations.
  5. 5Implement fairness-aware training and evaluation pipelines to ensure equitable model performance across groups.

Who benefits

AI DevelopmentSocial MediaRetailHealthcareGovernment

Key takeaways

  • Counterfactual fairness does not guarantee group fairness in image classification.
  • Latent attributes can explain the divergence between CF and GF in image tasks.
  • New datasets were created to enable simultaneous evaluation of both fairness types.
  • Counterfactual Knowledge Distillation (CKD) can help achieve group fairness by reducing reliance on latent attributes.

Original post by Sangwon Jung, Sumin Yu, Sanghyuk Chun, Taesup Moon

"arXiv:2607.06603v1 Announce Type: cross Abstract: The notion of algorithmic fairness has been actively explored from various aspects of fairness, such as counterfactual fairness (CF) and group fairness (GF). However, the exact relationship between CF and GF remains to be unclear,…"

View on X

Originally posted by Sangwon Jung, Sumin Yu, Sanghyuk Chun, Taesup Moon on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses

More in AI Engineering & DevTools

AI ResearchAI Engineering & DevTools

Transformers Learn Non-Invertible Modular Multiplication via Stratified Fourier Mechanisms.

This research investigates how small transformers learn modular integer multiplication over composite moduli, a non-invertible operation. It proposes the "monoid extension" theory, suggesting models partition input space into hierarchical algebraic regions where Fourier mechanisms apply, explaining how embeddings, attention, and local features contribute to the computation.

Zitong Andrew Chen, Junaid Hasan, Akhil Srinivasan, Hemkesh Bandi, Jarod AlperJul 9, 2026
AI Engineering & DevToolsAI Research

New Interpretable Model Handles Feature Interactions in Tabular Data.

This paper introduces Interaction Aware Interpretable Machine Learning (IAIML), a framework for tabular data that addresses the limitation of traditional interpretable models in capturing feature interactions. IAIML uses adaptive discretization, pairwise interaction scoring, and a partitioned explanation budget to achieve high accuracy while maintaining interpretability.

Srikumar KrishnamoorthyJul 9, 2026
AI ResearchAI Engineering & DevTools

Principles of Deep Feedforward ReLU Networks Unveiled.

This paper systematically studies the mechanisms of deep feedforward ReLU networks, generalizing principles from two-layer networks to deeper architectures. It explains how hidden-layer units form piecewise linear manifolds to divide input space and how paths and their relationships are central to understanding the back-propagation training solution.

Changcun HuangJul 9, 2026