AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

PRISM: Synergizing Vision Foundation Models via Self-organized Expert Specialization

Ying Tang, Dong Li, Youjia Zhang, Zikai Song, Junqing Yu, Wei Yang

ICML 2026 · Vision-language and multimodal › Human and embodied modeling · approach: Vision-language pretraining · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
replace dense multi-teacher distillation by dual-stream MoE (PRISM) over SAKPRISM ViT-B · PASCAL-Contextdelta_m (%): 0.83 -> 2.29Improved
PRISM vs SAK on indoor scenesPRISM ViT-B · NYUD-v2delta_m (%): 11.11 -> 10.59Mixed
scale student from ViT-B to ViT-LPRISM · PASCAL-Contextdelta_m (%): SAK 2.30 -> PRISM 3.16Improved
add MoE to dense baseline (ablation)PRISM ViT-S · PASCAL-Context (ablation)Semseg mIoU: 78.04 -> 79.19Improved
add Universal Anchor stream (ablation)PRISM ViT-S · PASCAL-Context (ablation)Normal mErr: 14.65 -> 14.28Improved
use FiLM-based router instead of task embedding concatenation (ablation)PRISM ViT-S · PASCAL-Context (ablation)Semseg mIoU: 78.87 -> 79.19Improved
add locality-aware decorrelation loss (ablation)PRISM ViT-S · PASCAL-Context (ablation)Parsing mIoU: 61.80 -> 69.25Improved
add locality-aware decorrelation loss (ablation)PRISM ViT-S · PASCAL-Context (ablation)Semseg mIoU: 79.80 -> 79.19Hurt
PRISM vs iso-FLOPs Wide ViT-S (ablation)PRISM ViT-S · PASCAL-Context (ablation)Semseg mIoU: 78.15 -> 79.19Improved
PRISM vs vanilla sparse MoE (ablation)PRISM ViT-S · PASCAL-Context (ablation)Semseg mIoU: 72.96 -> 79.19Improved
reduce expert MLP ratio to 1.0 (PRISM-lite) vs SAK ViT-SPRISM-lite ViT-S · PASCAL-Contextdelta_m (%): 0.43 -> 0.61Improved
sparse MoE dispatch vs SAK ViT-SPRISM-lite · EfficiencyLatency (ms): 20.25 -> 47.51Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy