HCFSLN: Adaptive Hyperbolic Few-Shot Learning for Multimodal Anxiety Detection

2025-11-10

AI Summary

To address the clinical dependency and scarcity of multimodal data in anxiety disorder diagnosis—which severely limit the generalizability of machine learning models—this paper proposes the first adaptive hyperbolic-space multimodal fusion framework for few-shot anxiety recognition. Methodologically, it pioneers the integration of hyperbolic geometry into mental health computing: hyperbolic embedding enhances inter-class separability, while cross-modal attention coupled with an adaptive gating mechanism enables dynamic alignment and fusion of speech, physiological signals, and video features. Evaluated on a newly constructed multimodal dataset comprising 108 participants, the framework achieves 88% accuracy using only a small number of labeled samples—surpassing the best baseline by 14 percentage points. This demonstrates the efficacy and generalization advantage of hyperbolic representation learning for few-shot psychiatric disorder identification.

Technology Category

Machine Learning: Multimodal LearningComputer Vision: Multi-modal VisionIntelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

User Modeling, Personalization and Recommendation: Psychology-informed user models and recommender systemsWeb Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web dataGraph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs

Abstract

Anxiety disorders impact millions globally, yet traditional diagnosis relies on clinical interviews, while machine learning models struggle with overfitting due to limited data. Large-scale data collection remains costly and time-consuming, restricting accessibility. To address this, we introduce the Hyperbolic Curvature Few-Shot Learning Network (HCFSLN), a novel Few-Shot Learning (FSL) framework for multimodal anxiety detection, integrating speech, physiological signals, and video data. HCFSLN enhances feature separability through hyperbolic embeddings, cross-modal attention, and an adaptive gating network, enabling robust classification with minimal data. We collected a multimodal anxiety dataset from 108 participants and benchmarked HCFSLN against six FSL baselines, achieving 88% accuracy, outperforming the best baseline by 14%. These results highlight the effectiveness of hyperbolic space for modeling anxiety-related speech patterns and demonstrate FSL's potential for anxiety classification.

Problem

• Detects anxiety disorders using limited multimodal data efficiently
• Addresses machine learning overfitting in small anxiety datasets
• Enables robust classification with speech, physiological and video inputs

Innovation

• Hyperbolic embeddings enhance feature separability
• Cross-modal attention integrates speech and video data
• Adaptive gating network enables robust few-shot classification