SPIN Unprocessed July 9, 2026 ai_technology research
STST-JEPA: Shallow-Target Spatio-Temporal Joint Embedding Prediction Architecture For EEG Self-Supervised Learning
View original on arxiv.orgOverview
arXiv:2607.06629v1 Announce Type: new Abstract: Brain age -- the age inferred from a physiological recording -- is an emerging biomarker whose deviation from chronological age tracks neurological and psychiatric burden, and EEG is an attractive substrate for it because it is cheap, portable, and temporally rich. Yet EEG brain-age models must contend with cross-site montage heterogeneity, small labelled cohorts, and dominant subject-level non-stationarity, and few EEG foundation models have been
SpinGraph analysis pending — check back after processing.
Ask AI about this story
Opens with the SpinGraph .md URL and structured context — one click, prompt included.
More from arXiv Machine Learning
View all →- At-Grok Is Not Converged:A Measurement-Validity Audit for Grokking Representation Metrics
- UASPL: Uncertainty-Aware Self-Paced Learning with Evidential Neural Networks
- Optimized Instance Alteration for Explaining and Assessing Robustness of Classifiers
- When Do Geometric Algebra Layers Beat Scalarization? A Controlled Study on SO(3)-Equivariant Vector Laws
- Does Demand Response Increase Vulnerability to Cyber Attacks by Adversarial Data Modifications?
- When Certificates Fail: A Unified Safety Framework for Embedded Neural Interface Models
Markdown (.md) · JSON-LD schema (.json) · Machine-readable for AI & GEO