DANCE: Detect and Classify Events in EEG

2026-05-11

AI Summary

Detecting and classifying neural events in continuous electroencephalography (EEG) signals without annotated event onsets remains a significant challenge. This work addresses this problem by formulating asynchronous neural decoding as a set prediction task and introduces an end-to-end deep learning architecture that jointly detects and classifies multi-type, multi-scale neural events directly from raw, unaligned EEG signals—eliminating the need for precise event onset annotations. Evaluated across ten diverse datasets spanning cognitive neuroscience, clinical applications, and brain–computer interfaces (BCIs), the proposed method substantially outperforms existing approaches. It achieves a new state-of-the-art performance in epileptic seizure monitoring and matches the accuracy of models that rely on exact event timing in BCI tasks.

Abstract

Event identification in continuous neural recordings is a critical task in neuroscience. Decoding in EEG is dominated by classifying windows aligned to known event onsets. However, while available in controlled experiments, such onsets are absent in continuous real-world monitoring. Here, we introduce DANCE, a deep learning pipeline that frames neural decoding as a set-prediction problem and jointly detects and classifies events directly from raw, unaligned signals. Evaluated separately on ten datasets curated from the literature with a wide variety of event types (ranging from milliseconds to minutes in duration), our model outperforms existing methods on a broad range of cognitive, clinical and BCI tasks. This single architecture establishes a new state of the art in the competitive task of seizure monitoring and matches the accuracy of onset-informed models for BCI tasks. Overall, our method marks a step towards end-to-end asynchronous neural decoding models

Problem

• EEG
• event detection
• event classification
• continuous neural decoding
• asynchronous decoding

Innovation

• DANCE
• event detection
• neural decoding
• EEG
• set prediction