Research map / Vision-language and multimodal / Human and embodied modeling
Learning Brain Representation with Hierarchical Visual Embeddings
Jiawen Zheng, Haonan Jia, MING LI, Yuhui Zheng, Yufeng Zeng, Yang Gao, Chen Liang
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 |
|---|---|---|---|
| use hierarchical visual fusion with MBP over best prior method | UBP baseline · THINGS-EEG intra-subject | top-1/top-5: 50.9/79.7 -> 75.7/94.6 | Improved |
| same, inter-subject EEG | UBP baseline · THINGS-EEG inter-subject | top-1/top-5: 12.4/33.4 -> 20.0/44.1 | Improved |
| same, MEG intra-subject | UBP baseline · THINGS-MEG intra-subject | top-1/top-5: 26.7/55.2 -> 33.7/60.5 | Improved |
| same, MEG inter-subject | UBP baseline · THINGS-MEG inter-subject | top-1/top-5: 2.2/10.4 -> 5.4/15.2 | Improved |
| add B32 to RN50 visual encoder (ablation) | fusion retrieval · THINGS-EEG intra-subject | top-1: 48.1 -> 56.9 | Improved |
| add VAE latent to B32 (ablation) | fusion retrieval · THINGS-EEG intra-subject | top-1: 52.2 -> 73.6 | Improved |
| add RN50 to B32+VAE (full triple) (ablation) | fusion retrieval · THINGS-EEG intra-subject | top-1: 73.6 -> 75.7 | Improved |
| add VAE to RN50 (ablation) | fusion retrieval · THINGS-EEG inter-subject | top-1: 12.7 -> 17.4 | Improved |
| add B32 to H14 fusion prior (ablation) | reconstruction prior · EEG reconstruction | CLIP: 0.773 -> 0.814 | Improved |
| add VAE to H14 fusion prior (ablation) | reconstruction prior · EEG reconstruction | CLIP: 0.773 -> 0.721 | Hurt |
| add VAE to H14+B32 fusion prior (ablation) | reconstruction prior · EEG reconstruction | PixCorr: 0.187 -> 0.195 | Mixed |
| use MBP instead of BrainProjection brain encoder (ablation) | retrieval with full fusion · THINGS-EEG intra-subject | top-1: 72.9 -> 75.7 | Improved |
Closest papers by meaning
- Human-Aligned Image Models Improve Visual Decoding from the Brain (ICML 2025 · similarity 0.91)
- What Does the Brain See? Multiview Neural Representation to Demystify the Brain-Visual Alignment (NeurIPS 2026 · similarity 0.89)
- MindAlign: Bridging EEG, Vision, and Language for Zero-Shot Visual Decoding (NeurIPS 2026 · similarity 0.89)
- Multigranular Evaluation for Brain Visual Decoding (AAAI 2026 · similarity 0.89)
- Brain decoding: toward real-time reconstruction of visual perception (ICLR 2024 · similarity 0.88)
- Wills Aligner: Multi-Subject Collaborative Brain Visual Decoding (AAAI 2025 · similarity 0.88)
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.
