Research map / Large language models / Language, safety and interpretability
DeWave: Discrete Encoding of EEG Waves for EEG to Text Translation
Yiqun Duan, Charles Zhou, Zhen Wang, Yu-Kai Wang, Chin-teng Lin
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| remove self-supervised codex pre-training (ablation) | DeWave · ZuCo word-level EEG features | BLEU-1/2/3: 41.35 -> 40.71 (BLEU-1) | Improved |
| add self-supervised codex pre-training (ablation) | DeWave · ZuCo word-level EEG features | ROUGE-1 F: 30.69 vs 30.63 | No clear effect |
| add self-supervised codex pre-training (ablation) | DeWave · ZuCo raw EEG waves | BLEU-1: 16.58 -> 20.51 | Improved |
| add self-supervised codex pre-training (ablation) | DeWave · ZuCo raw EEG waves | ROUGE-1 F: 18.33 -> 24.27 | Improved |
| increase codex size from 1024 to 8192 (ablation) | DeWave · ZuCo raw EEG waves | BLEU | Mixed |
| increase codex size from 1024 to 2048, then larger (ablation) | DeWave · ZuCo raw EEG waves | BLEU | Mixed |
| change latent dimension of the codex (ablation) | DeWave · ZuCo raw EEG waves | BLEU/ROUGE | No clear effect |
| perception field smaller than 80ms or larger than 240ms (ablation) | DeWave · ZuCo raw EEG waves | BLEU | Hurt |
| perception field 120ms to 240ms (ablation) | DeWave · ZuCo raw EEG waves | BLEU | No clear effect |
| add contrastive loss to the encoder | DeWave · ZuCo raw EEG waves | BLEU-1: 20.51 -> 21.09 | Improved |
| discrete codex representation vs EEG-to-Text baseline | DeWave · ZuCo word-level EEG features | BLEU-1: 40.12 -> 41.35 | Improved |
| discrete codex representation vs EEG-to-Text baseline | DeWave · ZuCo word-level EEG features | BLEU-4: 6.80 -> 8.22 | Improved |
Closest papers by meaning
- MindAlign: Bridging EEG, Vision, and Language for Zero-Shot Visual Decoding (NeurIPS 2026 · similarity 0.89)
- LEAF: Language-EEG Aligned Foundation Model for Brain-Computer Interfaces (NeurIPS 2026 · similarity 0.88)
- BrainOmni: A Brain Foundation Model for Unified EEG and MEG Signals (NeurIPS 2025 · similarity 0.87)
- NeuroLM: A Universal Multi-task Foundation Model for Bridging the Gap between Language and EEG Signals (ICLR 2025 · similarity 0.87)
- EEG-Language Pretraining for Highly Label-Efficient Clinical Phenotyping (ICML 2025 · similarity 0.87)
- SENSE: Semantic Neural Speech Synthesis from Brain Dynamics via Spatial Graph Encoding (NeurIPS 2026 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
