Research map / Label-efficient and robust learning / Self-supervised and contrastive
RECTOR: Masked Region-Channel-Temporal Modeling for Affective and Cognitive Representation Learning
Jinhan Liu, Mahsa Shoaran
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 |
|---|---|---|---|
| add masked predictive input reconstruction loss (ablation) | RECTOR · SEED (avg) | Weighted-F1: 71.6 -> 73.1 | Improved |
| add representation alignment loss (ablation) | RECTOR · SEED (avg) | Weighted-F1: 71.4 -> 73.1 | Improved |
| add topological structure modeling loss (ablation) | RECTOR · SEED (avg) | Weighted-F1: 71.1 -> 73.1 | Improved |
| add cross-view consistency loss (ablation) | RECTOR · SEED (avg) | Weighted-F1: 71.9 -> 73.1 | Improved |
| add adaptive functional partitioning (ablation) | RECTOR · SEED (avg) | Weighted-F1: 71.6 -> 73.1 | Improved |
| add top-p gating (ablation) | RECTOR · SEED (avg) | Weighted-F1: 72.1 -> 73.1 | Improved |
| use full MTRL objective instead of MAE-style random masking (ablation) | RECTOR · SEED (avg) | Weighted-F1: 70.3 -> 73.1 | Improved |
| use block-sparse RECTOR-SA instead of dense spatio-temporal attention (ablation) | RECTOR · SEED (avg) | Weighted-F1: 70.2 -> 73.1 | Improved |
| use block-sparse RECTOR-SA instead of dense attention (ablation) | RECTOR · MSIT | AUROC: 89.0 -> 92.1 | Improved |
| initialize AFP with random 9-region partition instead of anatomy (ablation) | RECTOR · SEED subject-dependent | Weighted-F1: 85.0 -> 80.9 | Hurt |
| scale pre-training data and model size (RECTOR-L+) | RECTOR · SEED, SEED-IV, DEAP | downstream performance | Improved |
| use learned AFP regions instead of fixed anatomical regions under cross-montage transfer (ablation) | RECTOR · DEAP to SEED, MSIT to ECR | transfer performance | Improved |
Closest papers by meaning
- CSBrain: A Cross-scale Spatiotemporal Brain Foundation Model for EEG Decoding (NeurIPS 2025 · similarity 0.87)
- Multi-dataset Joint Pre-training of Emotional EEG Enables Generalizable Affective Computing (NeurIPS 2025 · similarity 0.86)
- THD-BAR: Topology Hierarchical Derived Brain Autoregressive Modeling for EEG Generic Representations (NeurIPS 2025 · similarity 0.86)
- EEGPT: Pretrained Transformer for Universal and Reliable Representation of EEG Signals (NeurIPS 2024 · similarity 0.86)
- LUNA: Efficient and Topology-Agnostic Foundation Model for EEG Signal Analysis (NeurIPS 2025 · similarity 0.86)
- DMMR: Cross-Subject Domain Generalization for EEG-Based Emotion Recognition via Denoising Mixed Mutual Reconstruction (AAAI 2024 · 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.
