Scholarlink

With the rapid development of artificial intelligence and brain science, decoding text from brain signals for cognitive brain-computer interface (BCI) has garnered increasing attention. Among various brain signals, electroencephalography (EEG), a non-invasive modality, offers advantages such as ease of acquisition and high temporal resolution. This paper proposes a method for EEG signal acquisition based on control question stimulation with 8 electrode channels and introduces the ChiCo dataset, which aligns 1040 Chinese text corpora with EEG data. Furthermore, we present a novel framework, EEGTCT, based on the BART model and discrete codex coding, to decode EEG into corresponding Chinese text, achieving a BLEU-1 score of 52.95 on the ChiCo dataset. The findings provide valuable insights for improving BCI for individuals with speech disorders and for understanding the neural mechanisms of Chinese language processing.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

EEGTCT: Electroencephalogram-Based Chinese Text Decoding

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Sibo Zhang,
  • Bihui Yu,
  • Lili Zhou,
  • Gang Wang,
  • Jingxuan Wei,
  • Linzhuang Sun,
  • Liping Bu

摘要

With the rapid development of artificial intelligence and brain science, decoding text from brain signals for cognitive brain-computer interface (BCI) has garnered increasing attention. Among various brain signals, electroencephalography (EEG), a non-invasive modality, offers advantages such as ease of acquisition and high temporal resolution. This paper proposes a method for EEG signal acquisition based on control question stimulation with 8 electrode channels and introduces the ChiCo dataset, which aligns 1040 Chinese text corpora with EEG data. Furthermore, we present a novel framework, EEGTCT, based on the BART model and discrete codex coding, to decode EEG into corresponding Chinese text, achieving a BLEU-1 score of 52.95 on the ChiCo dataset. The findings provide valuable insights for improving BCI for individuals with speech disorders and for understanding the neural mechanisms of Chinese language processing.