Towards Counterfactual Fairness-aware Domain Generalization in Changing Environments

TL;DR

This work tackles domain generalization under evolving environments while preserving counterfactual fairness. It introduces DCFDG, a causal-disentangled framework that partitions exogenous information into four latent variables $U_s$, $U_{ns}$, $U_{v1}$, and $U_{v2}$, with temporal priors to capture domain evolution. The model optimizes a joint objective combining an ELBO-based reconstruction term, a counterfactual fairness loss $\mathcal{L}_{f}$, and an adversarial loss $\mathcal{L}_{TC}$ to disentangle semantic information from sensitive and environment-specific factors, with a theoretical KL-bound supporting cross-domain training. Empirical results on synthetic FairCircle and real-world Adult and Chicago Crime datasets show improved accuracy and stronger fairness guarantees across unseen domain sequences, highlighting the practical impact for fair, robust learning in changing environments.

Abstract

Recognizing the prevalence of domain shift as a common challenge in machine learning, various domain generalization (DG) techniques have been developed to enhance the performance of machine learning systems when dealing with out-of-distribution (OOD) data. Furthermore, in real-world scenarios, data distributions can gradually change across a sequence of sequential domains. While current methodologies primarily focus on improving model effectiveness within these new domains, they often overlook fairness issues throughout the learning process. In response, we introduce an innovative framework called Counterfactual Fairness-Aware Domain Generalization with Sequential Autoencoder (CDSAE). This approach effectively separates environmental information and sensitive attributes from the embedded representation of classification features. This concurrent separation not only greatly improves model generalization across diverse and unfamiliar domains but also effectively addresses challenges related to unfair classification. Our strategy is rooted in the principles of causal inference to tackle these dual issues. To examine the intricate relationship between semantic information, sensitive attributes, and environmental cues, we systematically categorize exogenous uncertainty factors into four latent variables: 1) semantic information influenced by sensitive attributes, 2) semantic information unaffected by sensitive attributes, 3) environmental cues influenced by sensitive attributes, and 4) environmental cues unaffected by sensitive attributes. By incorporating fairness regularization, we exclusively employ semantic information for classification purposes. Empirical validation on synthetic and real-world datasets substantiates the effectiveness of our approach, demonstrating improved accuracy levels while ensuring the preservation of fairness in the evolving landscape of continuous domains.

Figures (7)

• Figure 1: Causal Structure of DCFDG. The figure depicts the causal structures across two consecutive domains, wherein, due to the gradual evolution of the environment, we posit a correlation between the environmental information of each domain and that of the preceding domain.
• Figure 2: Network Architecture of DCFDG. We separately decouple the environmental information $U_{v1}$ and $U_{v2}$ for $X_s$ and $X_{ns}$, and employ the adversarial loss (Section \ref{['sec:dis_loss']}) to remove sensitive information from $U_s$. Semantic information $U_s$ and $U_{ns}$ are used for classification.
• Figure 3: Accuracy and total causal effect for each testing domain. The 1st, 3rd, and 5th figures illustrate the accuracy curves, while the 2nd, 4th, and 6th figures depict the total causal effect curves.
• Figure 4: Fairness-accuracy Trade-off on Adult and Crime. Each baseline is represented by five data points, corresponding to the outcomes under five distinct fairness parameter $\lambda_f$.
• Figure 5: Visualization of Fair-circle dataset. From left to right, these are respectively the training set, validation set, and test set.

Theorems & Definitions (7)

• Lemma 1
• Theorem 1
• Definition 1: Total Causal Effect (TCE) pearl2009causality
• Definition 2: Counterfactual Effect (CE) shpitser2008complete
• Lemma 2
• Definition 1
• Theorem 2