Uncovering Latent Depression Severity for Binary Depression Detection via Advantage-weighting Ranking

2026-07-07

AI Summary

This study addresses the challenges of severe multimodal feature distribution overlap and ambiguous decision boundaries in automatic depression detection by proposing a fine-grained multimodal fusion framework. The approach integrates temporal encoders with a cross-modal interaction module based on mutual-attention Transformers to enable deep inter-modality alignment. To further enhance discriminability, a binary advantage-weighted ranking loss is introduced, which dynamically identifies and reweights hard sample pairs while simultaneously reducing intra-class variance to reconstruct the underlying ordinal structure of depression severity. Experimental results on the D-Vlog and LMVD datasets demonstrate that the proposed method significantly improves binary depression detection accuracy, outperforming current state-of-the-art approaches.

Abstract

Automatic depression detection using audio-visual data faces significant challenges, particularly in disentangling overlapping feature distributions and establishing robust decision boundaries. To address this, we propose a fine-grained multimodal framework featuring a temporal encoder and a mutual transformer to facilitate deep cross-modal fusion. Our core contribution is the Binary Advantage-weighting Ranking Loss, which optimizes the latent space distribution through two complementary mechanisms: Advantage-weighted Separation, which mines hard pairs by computing a pairwise prediction difference matrix and dynamically weighting them based on their difficulty; and Advantage-weighted Compactness, which minimizes intra-class variance to force features to cluster around their respective class centers. Extensive experiments on D-vlog and LMVD demonstrate that our model reconstructs the latent ordinal structure by prioritizing hard pairs, thereby achieving state-of-the-art performance.

Problem

• depression detection
• feature distribution disentanglement
• decision boundary
• latent severity
• binary classification

Innovation

• Advantage-weighted Ranking
• Multimodal Fusion
• Hard Pair Mining
• Latent Ordinal Structure
• Binary Depression Detection