Joint-Embedding Predictive Architecture

Also known as JEPA

Overview

Joint-Embedding Predictive Architectures learn useful representations by predicting a target encoder’s latent features from visible or past context, rather than reconstructing every input detail. Common approaches use masked context-target prediction, online or EMA target encoders, and regularizers such as SIGReg to prevent collapse; action-conditioned variants model latent dynamics for planning. These methods are being applied to images, video, robotics, medical and scientific data, and time series, with evaluations ranging from linear probes to control and forecasting benchmarks. A central open tension is that predictable, non-collapsed features are not necessarily the features needed for a task: work tests how to preserve control-relevant or diagnostic information and how well prediction metrics reflect downstream performance.

Latest papers

CardsList

279 papers45 papers in the last four weeks, down 31% on the four weeks before. 0.3% of all new papers.

  1. JEPA as a Transferable Optimizer: Learning and Understanding Black-Box Optimization from Reward-Agnostic Pretraining

    Nov 17, 2026Mayukh Debnath, Manan Roy ChoudhuryLearning to OptimizeBlack-Box Optimization

  2. JEPA for astronomy: multi-survey spectral alignment puts stellar parameters on the leading principal axes

    Nov 17, 2026Xiaosheng Zhao, Rosemary F. G. Wyse, Yuan-Sen Ting +1Latent Space AlignmentJoint-Embedding Predictive Architecture

  3. CascadeJEPA: Enhancing 3D Geometric Representations via Spatially Coupled Cascaded Prediction

    Nov 17, 2026Kailai Zhou, Ziqi Chen, Xiaoning Xu +8Joint-Embedding Predictive Architecture3D Representation Learning

  4. SADL: On the Demand-Supply Geometry of Representation Collapse

    Nov 16, 2026Junhan Sun, Yigui Feng, Mingzhe Liu +7Representation CollapsePredictive Representation Learning

  5. PerturbJEPA: Joint-Embedding Predictive Learning of Cellular Perturbation Responses

    Nov 16, 2026Jiangshuan Pang, Youzhe He, Yiting Deng +5Single-Cell Perturbation ModelingJoint-Embedding Predictive Architecture

  6. JEPA4R: Embodied JEPA World Models with 4D Point Planning

    Nov 15, 2026Jiarui Dong, Zhiheng Liang, Yanjun Li +4World Models for RoboticsJoint-Embedding Predictive Architecture

  7. WA-JEPA: Goal-Aware Predictive Feedback Action Refinement for a World-Action Model

    Nov 14, 2026Tao An, Weihang Liang, Tianqi Wang +7Action-Conditioned World ModelsEfficient Inference for World Action Models

  8. Cracks in Embeddings: Robustness and Stability of JEPA-Based Chest X-Ray Representations

    Nov 13, 2026Zahid Mansur, Kalpana Yadav, Deepak MishraMedical Image Representation LearningAdversarial Robustness

  9. ActWorld-JEPA: Action-Conditioned JEPA World Models Meet Latent Planning for End-to-End Autonomous Driving

    Nov 12, 2026Ziying Song, Shengkai Zhang, Caiyan Jia +10World Models for Autonomous DrivingAutonomous Driving Planning

  10. Context-Predictive Low-Rank Adaptation for Face Presentation Attack Detection

    Nov 12, 2026Aminu Yusuf, Shakeel Muhammad Ibrahim, Sulaimon Oyeniyi Adebayo +3Face Presentation Attack DetectionVLM Adaptation

  11. DuoJEPA: Cross-Modal Translation Between Unaligned Self-Supervised Embedding Spaces

    Nov 11, 2026Mahmut S. Gokmen, Mitchell A. Klusty, Evan W. Damron +3Multimodal SSLCross-Modal Alignment

  12. Toward Cross-Rubric Surgical Skill Assessment via Joint Embedding Prediction

    Nov 11, 2026Jiajun Cheng, Haidar M. Abdul-Muhsin, Juha Choi +1Rubric-Based EvaluationSurgical Video Understanding

  13. RS-JEPA: Remote Sensing Joint-Embedding Predictive Architecture with Scan-Path Views for Cross-Sensor Image Retrieval

    Nov 10, 2026Tanishq Rachamalla, Srishti Kaushik, Muhammad Attique Khan +2Image RetrievalJoint-Embedding Predictive Architecture

  14. Market-Conditioned JEPA: Learning from Unpredictable Market Shocks

    Nov 10, 2026Yuqi Wang, Fengyuan Liu, Mukai Li +3Financial Time SeriesQuantitative Finance

  15. A Stochastic Latent Variable Generalization of Joint-Embedding Predictive Architectures

    Nov 6, 2026Nikos Giakoumoglou, Paschalis Giakoumoglou, Tania StathakiJoint-Embedding Predictive ArchitecturePredictive Representation Learning

  16. LEF-JEPA: Denoising-Trained Latent Energy Fields for Spatial Representation Learning

    Nov 6, 2026Zhiyuan Liu, Zhihao Xu, Sicheng Liu +1Representation LearningJoint-Embedding Predictive Architecture

  17. La-JEPA: Compact Robot Policies with Latent World Modeling

    Nov 5, 2026Mateo Clémente, George Z. Xue, Amin Abyaneh +5Robot Policy LearningWorld Action Models

  18. CAPE-JEPA - Cross-modal Adversarial Probabilistic Embedding JEPA

    Nov 4, 2026Aniket Junghare, Biplab BanerjeeCross-Modal Representation LearningCross-Modal Retrieval

  19. Cross-Domain Joint-Embedding Pretraining For Heterogeneous Industrial Time Series

    Nov 2, 2026Jonathan Adam Rico, Nagarajan Raghavan, Senthilnath JayaveluSelf-Supervised Time Series Representation LearningSelf-Supervised Pre-Training

  20. IsoAlign: Isoform-Aware DNA-Protein Alignment via Joint-Embedding Predictive Architectures

    Oct 30, 2026Niousha Sadjadi, Arman Seyed-Ahmadi, Tony Tu +8Genomic Foundation ModelsCross-Modal Alignment

  21. Latent Prediction Yields Cross-Subject Alignable Representations in Sensory-Evoked fMRI

    Oct 29, 2026Serwan Jassim, Victoria Bosch, Daniel Anthes +2fMRI Representation LearningfMRI-Based Visual Decoding