Fairness-Aware Few-Shot Learning for Audio-Visual Stress Detection

TL;DR

FairM2S tackles gender fairness in few-shot audio-visual stress detection by incorporating Equalized Odds constraints into both inner and outer meta-learning loops via adversarial gradient masking and gradient projection, using a BiLSTM-GRU backbone. The method optimizes a joint objective that combines classification, fairness, and regularization terms, and it releases SAVSD to support fairness research in low-resource settings. Empirical results across SAVSD, StressID, and AVD show that FairM2S achieves higher accuracy while substantially reducing fairness gaps (Eopp) and improving DI relative to strong baselines, with a favorable Pareto trade-off. These findings suggest FairM2S provides a practical, scalable path to equitable mental-health AI in real-world, data-scarce environments.

Abstract

Fairness in AI-driven stress detection is critical for equitable mental healthcare, yet existing models frequently exhibit gender bias, particularly in data-scarce scenarios. To address this, we propose FairM2S, a fairness-aware meta-learning framework for stress detection leveraging audio-visual data. FairM2S integrates Equalized Odds constraints during both meta-training and adaptation phases, employing adversarial gradient masking and fairness-constrained meta-updates to effectively mitigate bias. Evaluated against five state-of-the-art baselines, FairM2S achieves 78.1% accuracy while reducing the Equal Opportunity to 0.06, demonstrating substantial fairness gains. We also release SAVSD, a smartphone-captured dataset with gender annotations, designed to support fairness research in low-resource, real-world contexts. Together, these contributions position FairM2S as a state-of-the-art approach for equitable and scalable few-shot stress detection in mental health AI. We release our dataset and FairM2S publicly with this paper.

Figures (3)

• Figure 1: Pareto frontier comparison of models across datasets and shots. (Best viewed in color)
• Figure 2: Fairness sensitivity of FairM2S across datasets and hyperparameters for Eopp and DI. (Best viewed in color)
• Figure 3: Ablation results at 5-Shot across datasets.