OmniEEG-Bench: A Standardized Evaluation Benchmark for EEG Foundation Models
Researchers have introduced OmniEEG-Bench, a new standardized benchmark designed to evaluate foundation models for electroencephalography (EEG) data. This benchmark unifies 54 EEG datasets and organizes evaluation into six task families, addressing the fragmentation caused by inconsistent protocols and heterogeneous data. Initial benchmarking of 10 EEG foundation models revealed that both the diversity of pretraining data and model size are crucial for performance, indicating scaling-law behavior similar to large language models. AI
IMPACT Standardizes evaluation for EEG foundation models, potentially accelerating development and adoption in brain-computer interfaces.