Research map / Vision-language and multimodal / CLIP and zero-shot
Open Vocabulary Electroencephalography-to-Text Decoding and Zero-Shot Sentiment Classification
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Add diverse NR v1.0 training data to SR v1.0 (half1st) (ablation) | EEG-To-Text · SR v1.0 (half1st) + NR v1.0 (half1st) | BLEU-1: 33.5 -> 37.3 | Helped |
| Add NR v1.0 data from the second half of subjects (ablation) | EEG-To-Text · SR v1.0 (half1st) + NR v1.0 (half2nd) | BLEU-1: 33.5 -> 37.4 | Helped |
| Add NR v1.0 training data (ablation) | EEG-To-Text · SR v1.0 + NR v1.0 | BLEU-1: 34.5 -> 37.1 | Helped |
| Add TSR v1.0 training data (ablation) | EEG-To-Text · SR v1.0 + NR v1.0 + TSR v1.0 | BLEU-1: 37.1 -> 37.4 | No clear effect |
| Remove pretrained BART weights (ablation) | EEG-To-Text · SR v1.0 + NR v1.0 + NR v2.0 | BLEU-1: 40.1 -> 24.7 | Hurt |
| Use the zero-shot EEG-to-text and BART-classifier pipeline instead of a supervised EEG sentiment baseline | EEG-based sentence sentiment classification · SR v1.0 test | F1: 27.2 -> 55.6 | Helped |
| Use BART rather than BERT as the zero-shot pipeline classifier (ablation) | DECBART sentiment pipeline · SR v1.0 test | F1: 50.1 -> 55.6 | Helped |
| Remove pretrained weights from the EEG-to-text decoder (ablation) | DECBART + CLSBART · SR v1.0 test | F1: 55.6 -> 15.4 | Hurt |
| Add the multi-layer transformer encoder to the EEG-to-text decoder (ablation) | DECBART + CLSBART · SR v1.0 test | F1: 39.3 -> 55.6 | Helped |
| Decode EEG to text before sentiment classification rather than classify EEG features directly | EEG-based sentence sentiment classification · SR v1.0 test | F1: 27.2 -> 55.6 | Helped |
Closest papers by meaning
- EEG-Language Pretraining for Highly Label-Efficient Clinical Phenotyping (ICML 2025 · similarity 0.87)
- LibriBrain100: One Hundred Hours of Broad and Deep MEG Data for Neural Speech Decoding at Scale (NeurIPS 2026 · similarity 0.86)
- MindAlign: Bridging EEG, Vision, and Language for Zero-Shot Visual Decoding (NeurIPS 2026 · similarity 0.86)
- SENSE: Semantic Neural Speech Synthesis from Brain Dynamics via Spatial Graph Encoding (NeurIPS 2026 · similarity 0.86)
- Decoding Natural Images from EEG for Object Recognition (ICLR 2024 · similarity 0.86)
- Brain-OF: An Omnifunctional Foundation Model for fMRI, EEG and MEG (NeurIPS 2026 · similarity 0.85)
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.