Reimagining service delivery in the agentic era with Google Public Sector
cloud.google.com·1d ago
AI Research
arXiv CS.AI·Jun 11·5 min read
MA-DLE: Speech-based Automatic Depression Level Estimation via Memory Augmentation
TL;DR
Depression is often difficult to assess accurately, leading to a gap in effective mental health support. MA-DLE, a speech-based tool utilizing memory augmentation techniques, was developed to automatically estimate depression levels from audio data.
✦ Why It Matters
Engineers can leverage MA-DLE to develop innovative mental health assessment tools that utilize speech analysis.
Key Takeaways
- Explore memory augmentation for improved feature representation.
- Utilize GRU models with hierarchical attention for speech tasks.
- Evaluate your models on DAIC-WOZ and E-DAIC datasets.
- Focus on capturing long-term dependencies in temporal data.
How It Works
MA-DLE enhances GRU outputs by integrating two types of features: historical temporal features that provide context and dynamic memory features that reflect emotional changes. This selective integration reduces redundancy and improves the model's ability to capture relevant information over time.
The Hierarchical Attention Fusion module then combines these features, allowing the model to focus on the most informative aspects of the input data.