// ◐ World Models & Self-Supervised Learning
Joint-Embedding Predictive Architecture (JEPA)
Predicts masked content in representation space rather than pixels.
联合嵌入预测架构 JEPA 2023 Advanced core
IN PLAIN WORDS
LeCun's approach: predict target-block representations from context-block representations, ignoring unpredictable detail. I-JEPA for images, V-JEPA for video.
Used in: Meta I-JEPA / V-JEPA / V-JEPA 2 (video understanding, prediction and robot planning); PDE-JEPA (arXiv 2609.34715, Sep 2026) applies it to PDE dynamics
TIMELINE · prerequisites → this → what builds on it
01Original paper
Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecture
Mahmoud Assran, Quentin Duval, Ishan Misra, et al. · 2023
arXiv ↗ ✓ Verified · arXiv API
02Where it sits
Learn the left column first; the right column builds on this one.
Then you can learn
- Nothing in the atlas builds on it yet
Related:World Models
Left: prerequisites · right: builds on this · dashed: related. Full map →
03Learning resources
Official docs, explainers, courses and reference code. All links checked.
04Where the course teaches it
On learning paths:Generative Models & Frontier Research
For AI / data
- Verification: verified, checked 2026-10-21.
- JSON: /api/learn/algorithms.json (id:
jepa)