# A Contrastive Learning Based Convolutional Neural Network for ERP Brain-Computer Interfaces

Preprint · Yuntian Cui, Xinke Shen, Dan Zhang, Chen Yang

- Archive id: arxiv:2407.04738
- Published: 2024-07-02
- arXiv: https://arxiv.org/abs/2407.04738
- Source URL: https://arxiv.org/abs/2407.04738
- PDF: https://arxiv.org/pdf/2407.04738
- Provenance: arxiv

## Abstract

ERP-based EEG detection is gaining increasing attention in the field of brain-computer interfaces. However, due to the complexity of ERP signal components, their low signal-to-noise ratio, and significant inter-subject variability, cross-subject ERP signal detection has been challenging. The continuous advancement in deep learning has greatly contributed to addressing this issue. This brief proposes a contrastive learning training framework and an Inception module to extract multi-scale temporal and spatial features, representing the subject-invariant components of ERP signals. Specifically, a base encoder integrated with a linear Inception module and a nonlinear projector is used to project the raw data into latent space. By maximizing signal similarity under different targets, the inter-subject EEG signal differences in latent space are minimized. The extracted spatiotemporal features are then used for ERP target detection. The proposed algorithm achieved the best AUC performance in single-trial binary classification tasks on the P300 dataset and showed significant optimization in speller decoding tasks compared to existing algorithms.

_Abstract reproduced under CC0-1.0 (arXiv descriptive metadata)._

---
Space Kitty Archive · https://space-kitty.com · API: https://space-kitty.com/api/v1 · MCP: https://space-kitty.com/mcp · Premium: $0.20/request via x402 (Base Sepolia (testnet)). Metadata: arXiv (CC0), OpenAlex (CC0), Europe PMC (CC BY/CC0 works), ClinicalTrials.gov.