Research map / Vision-language and multimodal / Human and embodied modeling
BRACTIVE: A Brain Activation Approach to Human Visual Brain Learning
Xuan-Bac Nguyen, Hojin Jang, Xin Li, Samee U. Khan, Pawan Sinha, Khoa Luu
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 SOI features instead of CLS token features for the final layer (ablation) | BRACTIVE · NSD brain response prediction (P1) | Pearson correlation: 47.43 -> 50.31 | Improved |
| use SOI features instead of CLS token features for the final layer (ablation) | BRACTIVE · NSD brain response prediction (P7) | Pearson correlation: 47.31 -> 51.25 | Improved |
| pretrain the visual encoder with BRACTIVE (CLS) instead of ImageNet-1K | ViT-B-16 · NSD brain response prediction (P1) | Pearson correlation: 41.69 -> 47.43 | Improved |
| localize ROIs with BRACTIVE attention instead of GradCAM | GradCAM baseline · NSD human ROI localization (P1) | Dice score (%): 57.36 -> 68.98 | Improved |
| average attention maps across participants | GradCAM · NSD human ROI localization (average) | Dice score (%): 61.24 -> 54.84 | Hurt |
| average attention maps across participants | BRACTIVE · NSD human ROI localization (average) | Dice score (%): 69.70 -> 72.57 | Improved |
| train with BLIP-2 dense captions instead of COCO captions (ablation) | BRACTIVE · NSD human ROI localization (P1) | Dice score (%): 68.98 -> 69.10 | No clear effect |
| train with BLIP-2 dense captions instead of COCO captions (ablation) | BRACTIVE + SOI · NSD brain response prediction (P7) | Pearson correlation: 51.25 -> 51.72 | No clear effect |
| use BRACTIVE pretrained encoder instead of ImageNet-1K | ViT-B-16 Mask R-CNN · COCO detection | AP box: 49.8 -> 51.4 | Improved |
| use BRACTIVE pretrained encoder instead of CLIP | ViT-B-16 UperNet · ADE20K semantic segmentation | mIoU: 47.6 -> 48.7 | Improved |
| use BRACTIVE pretrained encoder instead of ImageNet-1K | ViT-B-16 Mask R-CNN · COCO instance segmentation | AP segm: 44.5 -> 45.7 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- GradCAM — Grad-CAM: Visual Explanations From Deep Networks via Gradient-Based Localization (ICCV 2017) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- BrainACTIV: Identifying visuo-semantic properties driving cortical selectivity using diffusion-based image manipulation (ICLR 2025 · similarity 0.88)
- A Convolutional Neural Network Interpretable Framework for Human Ventral Visual Pathway Representation (AAAI 2024 · similarity 0.86)
- In Silico Mapping of Visual Categorical Selectivity Across the Whole Brain (NeurIPS 2025 · similarity 0.86)
- Meta-Learning an In-Context Transformer Model of Human Higher Visual Cortex (NeurIPS 2025 · similarity 0.86)
- MoRE-Brain: Routed Mixture of Experts for Interpretable and Generalizable Cross-Subject fMRI Visual Decoding (NeurIPS 2025 · similarity 0.86)
- Toward Generalizing Visual Brain Decoding to Unseen Subjects (ICLR 2025 · similarity 0.86)
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.
