Orthogonal JEPA: Factorized Predictive States for Latent World Models

• URL: http://arxiv.org/abs/2608.20065v1
• Date: Thu, 20 Aug 2026 13:59:57 GMT
• Title: Orthogonal JEPA: Factorized Predictive States for Latent World Models
• Abstract summary: We introduce method, a latent world-modeling framework based on predictive factorization.<n>We show that method can allocate redundant capacity to dominant signals while providing weak or conflicting gradients to less dominant predictive structure.<n>Experiments on controlled vision, single-cell transcriptomics, longitudinal health records, continuous control, and molecular dynamics evaluate representation quality, forecasting, planning, and long-horizon stability.
• Score: 95.20158869896393
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: World models construct latent states that support prediction, planning, and reasoning about an underlying system. Joint-embedding predictive architectures (JEPAs) offer a direct way to learn such states by predicting targets in representation space instead of reconstructing every detail of the observation. Standard JEPAs, however, organize all predictable content through one target embedding and one prediction pathway. In complex systems, this monolithic state can allocate redundant capacity to dominant signals while providing weak or conflicting gradients to less dominant predictive structure. We introduce \method, a latent world-modeling framework based on orthogonal predictive factorization. Learned basis matrices analyze each target state into multiple components, and a dedicated prediction branch estimates each component from a shared context representation. Predictive regression preserves the factor magnitudes required for state synthesis, an orthogonality objective discourages repeated directions, factor-activity regularization maintains variation in projected targets, and online variance regularization discourages coordinate-wise encoder collapse. Predicted components are synthesized into a complete latent state that can be used by a readout, decoder, planner, or autoregressive rollout. The same predictive-state mechanism applies when the target is temporally future, spatially hidden, or another partial observation of the same system. Experiments on controlled vision, single-cell transcriptomics, longitudinal health records, continuous control, and molecular dynamics evaluate representation quality, forecasting, planning, and long-horizon stability.

Related papers

• UniJEPA: A Unified Joint-Embedding Predictive Architecture for Task-Agnostic Visual World Modeling [24.745715007832953]
  Joint-Embedding Predictive Architectures (JEPAs) have emerged as a principled framework for self-supervised learning of world models in compact latent spaces.<n>We present UniJEPA, a unified JEPA that jointly learns photometric prediction (image-level transformations) and temporal prediction (video-level next-state dynamics) in one shared latent space.
  arXiv  Detail & Related papers  (2026-08-07T16:55:58Z)
• APO: Unsupervised Atomic Policy Optimization for 3D Structure Prediction of Atomic Systems [64.56252173115246]
  We propose (Atomic Policy Optimization) a fully unsupervised alignment framework that eliminates the need for ground-truth reference structures.<n>APO adapts group-relative policy optimization to 3D atomic environments, utilizing a novel dual-reward mechanism.<n>Our results suggest that intrinsic physical consistency can serve as a superior guide for alignment compared to noisy, supervised coordinate matching.
  arXiv  Detail & Related papers  (2026-07-30T17:21:58Z)
• On the Identifiability of Controlled World Models [34.31084112465058]
  This paper presents a joint identifiability condition for controlled world models with Gaussian latent states.<n>We prove that when this condition holds, minimizing the LeJEPA-style predictive objective can recover both latent states and controlled dynamics.<n>We also characterize an attainable amplification of counterfactual prediction error that scales inversely with the weakest conditional action-excitation margin.
  arXiv  Detail & Related papers  (2026-07-24T15:49:12Z)
• A Bayesian Framework for Evaluating Scenario Compatibility in Generative Population Synthesis [6.719751155411076]
  We propose an ensemble-based Bayesian updating framework to quantify scenario compatibility in conditional population synthesis.<n>A population-aware conditional variational autoencoder is developed to learn a distribution over plausible population structures.<n>Experiments demonstrate that scenario impact depends not only on target magnitude but also on alignment with the learned joint structure.
  arXiv  Detail & Related papers  (2026-07-03T10:50:44Z)
• Learning Transferable Predictability Representations [2.1505812656718795]
  We study the problem of assigning a scalar score to a short trajectory window that reflects its position on an ordered continuum of predictability regimes.<n>We formalize this as ordinal estimation over a five-level predictability ladder and identify a structural source of cross-system ambiguity.<n>We propose the Gauge-Fixed Ordinal Network (GON), a temporal convolutional model trained with an anchor-and-variance objective that pins level-wise score means to shared target coordinates.
  arXiv  Detail & Related papers  (2026-05-28T21:38:49Z)
• ThinkJEPA: Empowering Latent World Models with Large Vision-Language Reasoning Model [53.15040805435013]
  Vision-language models (VLMs) provide strong semantic grounding and general knowledge by reasoning over uniformly sampled frames.<n>We propose a VLM-guided JEPA-style latent world modeling framework that combines dense-frame dynamics modeling with long-horizon semantic guidance.
  arXiv  Detail & Related papers  (2026-03-23T17:59:42Z)
• BiJEPA: Bi-directional Joint Embedding Predictive Architecture for Symmetric Representation Learning [0.0]
  BiJEPA enforces cycle-consistent predictability between data segments.<n>We evaluate BiJEPA on three distinct modalities: synthetic periodic signals, chaotic Lorenz attractor trajectories, and high-dimensional image data.
  arXiv  Detail & Related papers  (2026-02-10T01:16:20Z)
• Topology-Aware Conformal Prediction for Stream Networks [68.02503121089633]
  We propose Spatio-Temporal Adaptive Conformal Inference (textttCISTA), a novel framework that integrates network topology and temporal dynamics into the conformal prediction framework.<n>Our results show that textttCISTA effectively balances prediction efficiency and coverage, outperforming existing conformal prediction methods for stream networks.
  arXiv  Detail & Related papers  (2025-03-06T21:21:15Z)
• DR-Label: Improving GNN Models for Catalysis Systems by Label Deconstruction and Reconstruction [72.20024514713633]
  We present a novel graph neural network (GNN) supervision and prediction strategy DR-Label. The strategy enhances the supervision signal, reduces the multiplicity of solutions in edge representation, and encourages the model to provide node predictions robust. DR-Label was applied to three radically distinct models, each of which displayed consistent performance enhancements.
  arXiv  Detail & Related papers  (2023-03-06T04:01:28Z)
• Target-Embedding Autoencoders for Supervised Representation Learning [111.07204912245841]
  This paper analyzes a framework for improving generalization in a purely supervised setting, where the target space is high-dimensional. We motivate and formalize the general framework of target-embedding autoencoders (TEA) for supervised prediction, learning intermediate latent representations jointly optimized to be both predictable from features as well as predictive of targets.
  arXiv  Detail & Related papers  (2020-01-23T02:37:10Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.