Research map / Time series / Long-term forecasting
A foundation model with multi-variate parallel attention to generate neuronal activity
Francesco S. Carzaniga, Michael Hersche, Abu Sebastian, Kaspar Schindler, Abbas Rahimi
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 MVPA instead of vanilla attention (ablation) | MV-Llama · SWEC seizure detection | Cohen's Kappa: 0.11 -> 0.57 | Improved |
| add MVPA instead of vanilla attention (ablation) | MV-Llama · Brain TreeBank Pitch | accuracy: 0.63 -> 0.83 | Improved |
| scale MVPFormer-S to MVPFormer-M | MVPFormer-S · SWEC seizure detection | Cohen's Kappa: 0.57 -> 0.61 | Improved |
| add generative contrastive pre-training (ablation) | MVPFormer-S · SWEC seizure detection | Cohen's Kappa: 0.52 -> 0.54 | Improved |
| add MVPFormer over Brant-2 | Brant-2 · MAYO seizure detection | F1: 0.19 -> 0.36 | Improved |
| add MVPFormer over Brant-2 | Brant-2 · FNUSA seizure detection | F1: 0.46 -> 0.46 | No clear effect |
| use MVPFormer-S without electrode positions vs PopT with positions | PopT with positions · Brain TreeBank Speech | accuracy: 0.93 -> 0.90 | Hurt |
| add MVPA components (content, time, channel) versus none (ablation) | MVPA · Weather forecasting | MSE: 0.27 -> 0.25 | Improved |
| add MVPA components versus none (ablation) | MVPA · ETTh1 forecasting | MSE: 0.47 -> 0.45 | Improved |
| add time and channel components to content-only MVPA (ablation) | Content-only MVPA · ETTh2 forecasting | MSE: 0.38 -> 0.38 | No clear effect |
| use MVPFormer instead of vanilla Transformer | Transformer · ETTh2 forecasting | MSE: 3.37 -> 0.38 | Improved |
| use MVPFormer instead of PatchTST | PatchTST · ETTh1 forecasting | MSE: 0.45 -> 0.45 | No clear effect |
Closest papers by meaning
- DMNet: Self-comparison Driven Model for Subject-independent Seizure Detection (NeurIPS 2024 · similarity 0.86)
- CSBrain: A Cross-scale Spatiotemporal Brain Foundation Model for EEG Decoding (NeurIPS 2025 · similarity 0.85)
- NeuroAtlas: Benchmarking Foundation Models for Clinical EEG and Brain-Computer Interfaces (NeurIPS 2026 · similarity 0.85)
- BaRISTA: Brain Scale Informed Spatiotemporal Representation of Human Intracranial Neural Activity (NeurIPS 2025 · similarity 0.85)
- EEGPT: Pretrained Transformer for Universal and Reliable Representation of EEG Signals (NeurIPS 2024 · similarity 0.85)
- S$^2$M-Former: Spiking Symmetric Mixing Branchformer for Brain Auditory Attention Detection (NeurIPS 2025 · 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.
