aiqbench analysis
This research advances emotion recognition in wearable devices by leveraging self-supervised learning on graph neural networks to overcome data scarcity and labeling bias. By utilizing subgraph sampling with only 20–25% of available labels, the method achieves significant accuracy gains over full-resource baselines while maintaining robustness against inter-subject variability. The approach demonstrates how SSL-driven augmentation can effectively model complex emotional states like arousal and valence in real-world, unstructured environments.
Self-Supervised Graph Representation Learning for In-The-Wild Wearable and Smartphone based Emotion Recognition
arXiv:2608.22387v1 Announce Type: cross Abstract: Wearable and smartphone-based emotion recognition (WER) remains a challenging setting in affective computing, due to the notorious difficulty and bias associated with in-the-wild label collection. The high inter-and intra-subject emotional variability motivates us to explore WER modeling through graph node classification in a limited resources learning scheme powered by Self-Supervised Learning (SSL) graph masking augmentation tasks. We employ a subgraph sampling approach during training, utilizing labeled and unlabeled data, along with supervised, semi-supervised, and SSL mechanisms in a multi-task inductive graph neural network architecture. Our evaluations on K-EmoPhone through leave-one-group-out cross-validation in the binary arousal and valence tasks yield average accuracy gains of 4.3% and 7.8%, compared to the full resource setting, utilizing only 20% and 25% of the labels, respectively. Our model analysis sheds light on the relation of SSL graph augmentations to emotional arousal and valence and justifies the approach of SSL-driven subgraph training for in-the-wild WER.
Abstract:Wearable and smartphone-based emotion recognition (WER) remains a challenging setting in affective computing, due to the notorious difficulty and bias associated with in-the-wild label collection. The high inter-and intra-subject emotional variability motivates us to explore WER modeling through graph node classification in a limited resources learning scheme powered by Self-Supervised Learning (SSL) graph masking augmentation tasks. We employ a subgraph sampling approach during training, utilizing labeled and unlabeled data, along with supervised, semi-supervised, and SSL mechanisms in a multi-task inductive graph neural network architecture. Our evaluations on K-EmoPhone through leave-one-group-out cross-validation in the binary arousal and valence tasks yield average accuracy gains of 4.3% and 7.8%, compared to the full resource setting, utilizing only 20% and 25% of the labels, respectively. Our model analysis sheds light on the relation of SSL graph augmentations to emotional arousal and valence and justifies the approach of SSL-driven subgraph training for in-the-wild WER.
Published at ICASSP 2025. Copyright 2025 IEEE
Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Signal Processing (eess.SP)
arXiv:2608.22387 [cs.LG]
(or arXiv:2608.22387v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2608.22387 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
https://doi.org/10.1109/ICASSP49660.2025.10888648 Focus to learn more DOI(s) linking to related resources
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 · 266 words
Related coverage
Exploring Hierarchical Interest Representation For Meta Ads Deep Funnel Optimization
Meta Engineering · Jul 15, 2026
Complementary Supervised and Self-Supervised Representations for Out-of-Distribution Graph Learning
arXiv · Oct 7, 2026
Microsoft’s AI chief says superintelligence is near, but won’t take your job
The Verge · Jun 8, 2026
Frequency-Corrupt Based Graph Self-Supervised Learning
arXiv · Jul 21, 2026