aiqbench analysis
EviDep introduces a novel multimodal framework for depression severity estimation that explicitly quantifies aleatoric and epistemic uncertainty using evidential regression. By disentangling shared and modality-specific features across audio-visual inputs, the model improves prediction reliability and identifies high-error cases under data degradation. This work advances the application of probabilistic machine learning in sensitive health domains where confidence metrics are critical for clinical trust.
EviDep: Uncertainty-Aware Multimodal Depression Estimation via Disentangled Evidential Learning
arXiv:2604.16579v3 Announce Type: replace-cross Abstract: Audio--visual recordings provide complementary cues for estimating depression severity, but their informativeness varies across time and modalities. Point predictions alone do not express the uncertainty associated with these estimates. We present EviDep, a multimodal evidential regression framework that integrates multi-scale temporal modeling and shared--private representation learning for uncertainty-aware depression estimation. Frequency-aware Feature Extraction decomposes behavioral feature sequences into multiple frequency bands and refines them with scale-specific experts. Disentangled Evidential Learning encourages the disentanglement of cross-modal shared and modality-specific information in the refined features. Multi-branch Evidential Regression maps the resulting shared and private representations to three Normal-Inverse-Gamma (NIG) outputs and uses evidence-weighted aggregation to estimate depression severity and quantify aleatoric and epistemic uncertainty. Experiments on AVEC 2013, AVEC 2014, DAIC-WOZ, and E-DAIC show competitive prediction accuracy, with ablation studies supporting the contributions of frequency-aware refinement and shared--private disentanglement. Further analyses show that estimated epistemic uncertainty helps identify higher-error predictions, while both uncertainty estimates generally increase under controlled feature degradation.
Abstract:Audio--visual recordings provide complementary cues for estimating depression severity, but their informativeness varies across time and modalities. Point predictions alone do not express the uncertainty associated with these estimates. We present EviDep, a multimodal evidential regression framework that integrates multi-scale temporal modeling and shared--private representation learning for uncertainty-aware depression estimation. Frequency-aware Feature Extraction decomposes behavioral feature sequences into multiple frequency bands and refines them with scale-specific experts. Disentangled Evidential Learning encourages the disentanglement of cross-modal shared and modality-specific information in the refined features. Multi-branch Evidential Regression maps the resulting shared and private representations to three Normal-Inverse-Gamma (NIG) outputs and uses evidence-weighted aggregation to estimate depression severity and quantify aleatoric and epistemic uncertainty. Experiments on AVEC 2013, AVEC 2014, DAIC-WOZ, and E-DAIC show competitive prediction accuracy, with ablation studies supporting the contributions of frequency-aware refinement and shared--private disentanglement. Further analyses show that estimated epistemic uncertainty helps identify higher-error predictions, while both uncertainty estimates generally increase under controlled feature degradation.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2604.16579 [cs.LG]
(or arXiv:2604.16579v3 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2604.16579 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 253 words
Related coverage
Deep Evidential Regression for Sparse Forest Height Estimation from Multimodal Satellite Imagery
arXiv · Sep 15, 2026
MS-MEM: Multi-Skill Manipulation-Enhanced Mapping via Uncertainty- and Disturbance-Aware Action Selection
arXiv · Sep 3, 2026
HierarchicalDAEW: Domain-Aware Edge-Weighted Graph Convolution with Evidential Uncertainty for Multi-Section Spatial Gene Expression Prediction from H&E Histology
arXiv · Jul 24, 2026
Uncertainty-Aware Decision Making in Multimodal Large Language Models
arXiv · Aug 19, 2026