JEPA

Joint-embedding predictive architectures: non-generative predictive world models. (82 papers)

2026-11-13 · arXiv:2611.16182

Same-State Pairs Make One JEPA Proof of the Platonic Representation Hypothesis Vacuous

Xin Li

The authors show that, under the hard-whitening JEPA population objective with same-state positive pairs, the induced world kernel becomes the identity, so every centered whitened state function is a global minimizer and rotation-uniqueness becomes vacuous. They further show an orbit-level failure: two minimizers can reach the maximum best-rotation residual. One surviving result is a gap-free lower bound on alignment error, but it cannot certify good-view alignment in the degenerate regime.

JEPAAI summary
2026-11-13 · arXiv:2611.16639

Asymmetric Multimodal Learning with Decomposed JEPAs

Yusuf Shihata

The authors show that cross-modal alignment as a sole objective can lose task-critical private and synergistic information under informational asymmetry, leading to semantic collapse on compositional VQA. They prove an “Alignment Redundancy Bound” limiting aligned latents to the redundant information atom. They introduce Decomposed JEPA (D-JEPA) with a shared core plus private fibers, averting collapse and improving downstream accuracy.

JEPAAI summary
2026-11-13 · arXiv:2611.17146

HiVE-VQA: Hierarchical Interaction via Volumetric Embedding for Video Quality Assessment

Robbie Hamilton, Fan Zhang, David Bull

HiVE-VQA applies a JEPA-based frozen video encoder (V-JEPA 2.1 ViT-B) to no-reference video quality assessment. It extracts token grids from multiple depths, stacks them into a hierarchical spatio-temporal volume, and trains only a compact 3D convolutional head (5.2M parameters, 5.7% of total). Trained on LSVQ and evaluated without fine-tuning, it outperforms 17 SOTA methods on SROCC and PLCC across intra- and cross-dataset benchmarks.

JEPACode availableAI summary
2026-11-11 · arXiv:2611.13572

Market-Conditioned JEPA: Learning from Unpredictable Market Shocks

Yuqi Wang, Fengyuan Liu, Mukai Li et al.

The authors introduce Market-Conditioned JEPA (MC-JEPA), a joint-embedding predictive model for financial time series that conditions training prediction on a market factor constructed from other assets’ future embeddings. In a latent factor model, conditioning makes exposure necessary for optimal prediction and reduces noise in learning asset-specific signals. On 11 years of minute-level Chinese A-share data, MC-JEPA improves ridge-probe RankIC and trading backtest metrics over per-asset JEPA.

JEPAAI summary
2026-11-11 · arXiv:2611.14496

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

Tanishq Rachamalla, Srishti Kaushik, Muhammad Attique Khan et al.

RS-JEPA is a compact, label-free cross-sensor image retrieval method that uses a shared Vision Mamba encoder and scan-path views: two reading orders of the same image are treated as self-supervised views, predicting hidden patch features from visible ones. It adds Adversarial View Coupling (AVC) to make prediction harder and a neighbourhood consensus loss to align cross-sensor similarity geometry. The authors report the highest F1@5 among twelve re-implemented methods.

JEPACode availableAI summary
2026-11-10 · arXiv:2611.10658

The Delta Gap: Cellular World Models for Perturbation Response Prediction Are Far Harder Than Global Correlation Implies

Junhui Gao

The authors show that cellular perturbation world models are often evaluated with global Pearson correlation, which is dominated by non-responsive genes and inflates apparent performance. On Replogle K562, GEARS gets global Pearson 0.977 but delta Pearson only 0.076; a trivial mean-training-delta baseline gets 0.267. They propose delta-space evaluation and show delta-aligned training plus post-hoc corrections improve delta Pearson and direction/gene recovery, with zero-shot transfer across cell lines.

JEPAAI summary
2026-11-10 · arXiv:2611.12137

Good Views Are Not Enough: Global Counterexamples for Gaussian-Regularized JEPAs

Anran Xu

The authors show that Gaussian-regularized JEPA objectives can have global optima that do not identify a scene up to rotation. They give globally certified counterexamples where Gaussian marginal matching rewards retaining independently resampled observation variation, and this can outweigh its prediction cost, forcing nuisance retention and allowing equal-value minimizers outside a single rotation orbit.

JEPAAI summary
2026-11-05 · arXiv:2611.05570

ChangeJEPA: Predict Before You Compare for Remote Sensing Change Detection

Buddhi Wijenayake, Dhananjaya Jayasundara, Vishal M. Patel

ChangeJEPA reformulates remote sensing change detection as latent post-state prediction. Given a pre-change image and partial post-change evidence, it predicts a full post-change latent representation under two complementary conditions (prior vs updated). Change is inferred from disagreement among pre representation, observed post target, and predicted latent states, aiming to suppress nuisance variation while keeping semantic consistency.

JEPACode availableAI summary
2026-11-05 · arXiv:2611.06093

Drift-adaptive Representation Preservation for Domain-Incremental Learning

Hyunmin Kong, Gyutae Oh, Jitae Shin

The authors address domain-incremental learning forgetting by tracking per domain–class representation drift from a fixed class anchor. Drift is measured using a repurposed JEPA context-to-target predictor without storing past images (store compact context summaries in backbone feature space). Drift-adaptive consolidation and drift-weighted robust prototype construction allocate preservation by measured drift, improving accuracy–forgetting balance across DomainNet, Office-Home, CORe50, and CDDB-Hard with low training memory and smallest storage among compared methods.

JEPAAI summary
2026-11-05 · arXiv:2611.06247

CAPE-JEPA - Cross-modal Adversarial Probabilistic Embedding JEPA

Aniket Junghare, Biplab Banerjee

CAPE-JEPA is a probabilistic cross-modal JEPA for remote sensing cross-modal image retrieval. It replaces deterministic latent prediction with a cross-attentional Gaussian predictor (per-patch mean/variance) trained by negative log-likelihood, adds adversarial cross-modal distributional alignment via worst-case Epps–Pulley tests without explicit sample pairing, and uses in-training InfoNCE with attention-pooled heads for scene-level discrimination.

JEPAAI summary
2026-11-03 · arXiv:2611.00241

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

Niousha Sadjadi, Arman Seyed-Ahmadi, Tony Tu et al.

IsoAlign is a multimodal genomic pretraining framework for isoform-aware DNA-protein representations using two JEPA architectures: DP-JEPA (predict protein representations from DNA via contrastive alignment) and IM-JEPA (mask protein isoforms and predict them using DNA anchors and visible isoform context). The authors evaluate DNA representations on tasks including fitness, variant effect prediction, gene expression, and TF binding.

JEPAAI summary
2026-11-03 · arXiv:2611.02489

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

Jonathan Adam Rico, Nagarajan Raghavan, Senthilnath Jayavelu

The authors propose JEPA-HITS, a channel-independent joint-embedding predictive pretraining method for heterogeneous industrial time series. They learn one shared JEPA encoder across battery, CMP, CNC, and bearing data while preserving dataset-specific temporal scales, using leave-one-domain-out (LODO) pretraining to test transfer to unseen domains. Compared to the same architecture with random initialization, LODO pretraining reduces RMSE on NASA/CMP/CNC and improves bearing fault macro-F1.

JEPAAI summary
2026-10-30 · arXiv:2610.38445

DES-JEPA: Learning a Shared Vocabulary for High-Dimensional Discrete Event Sequences

Hugo Math, Maximilian Forstenh\"ausler, Christos Anagnostopoulos

DES-JEPA learns a shared discrete codebook for discrete event sequences with unaligned, domain-specific event vocabularies by training on synthetic stochastic processes. DES-PFN reuses the frozen codebook for in-context next-event prediction (zero-shot, no task-specific fine-tuning). The authors report competitive zero-shot performance on eight real datasets and scaling to 700 distinct event types.

JEPAAI summary
2026-10-30 · arXiv:2610.39302

Proving and Improving the Platonic Representation Hypothesis with JEPAs

Randall Balestriero, Yann LeCun

The authors prove a “Platonic Representation Theorem” for JEPAs: if modalities share the same positive-pair distribution formed from two observations of the same scene (“good views”), their learned representations match up to one rotation. They also show next-token prediction is a “bad-view” JEPA and validate the theory with scaling experiments from 145k captions to texts of 940M web images, reporting that good views are far more sample-efficient than next-token prediction and yield stronger cross-modal alignment.

JEPAAI summary
2026-10-30 · arXiv:2610.40280

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

Serwan Jassim, Victoria Bosch, Daniel Anthes et al.

The authors adapt JEPA to pre-train participant-specific fMRI encoders using brain-only data from NSD. Encoders never see stimulus images or stimulus-derived representations. They evaluate frozen encoders with MindEye2 for image reconstruction and CorText for brain captioning, including cross-participant generalization. The authors report that independently trained participant embeddings can be aligned fully unsupervised without shared stimuli, enabling cross-subject decoding without fine-tuning.

JEPAAI summary
2026-10-30 · arXiv:2610.40356

JEDI: Joint-Embedding Demonstration Compression for In-Context Imitation Learning

Jiazhao Liang, Zhewen He, Shuaihang Yuan et al.

JEDI compresses in-context imitation demonstrations into k latent slots trained to predict the query episode’s next-stage keyframe in a frozen video-JEPA (V-JEPA) space. The predictor selects which cached demonstrations to place in context. Post-training enables conditioning on human-hand demonstrations via keypoint tokens. Experiments on LIBERO and real robots show improved many-shot context performance and better success/demonstration following than action-only compression.

JEPAAI summary
2026-10-28 · arXiv:2610.36121

SuperJEPA: Pretraining with a Supervision-Augmented Joint-Embedding Predictive Architecture for High Energy Physics

Dylan Sheldon Rankin, Ho Fung Tsoi

SuperJEPA introduces a joint-embedding predictive architecture (JEPA) augmented with a supervised classification objective for high energy physics. The authors report improved downstream jet-tagging performance over supervised pretraining alone across multiple tasks, model sizes, and pretraining dataset sizes, with gains persisting for both frozen and fine-tuned encoders.

JEPAAI summary
2026-10-28 · arXiv:2610.36493

AC-JEPA: Anchored Cross-Modal Joint Embedding Predictive Architecture for Visual-to-Symbolic Translation

Chaofei Hong, Ziyang Chen, Baochen Wang et al.

AC-JEPA integrates cross-modal latent prediction into an SFT-trained image-to-SMILES VLM. It uses bidirectional masked-state JEPA in both visual and symbolic directions as an auxiliary objective while keeping a causal SFT “anchor” for token-level supervision. The authors report improved accuracy on WildMol Simple with clean-pair training and further gains with latent self-training using pseudo-labeled pairs.

JEPAAI summary
2026-10-28 · arXiv:2610.36662

SWIM: Learning What Matters from Behaviors for JEPA World Models

Jiaming Hu, Shi Bo, Yan Zheng et al.

SWIM learns which simulator state coordinates matter for JEPA-style planning from offline behavior. It trains a gated inverse-dynamics model to predict logged actions from current states and observed future displacements, using it to define a learned weighted state metric for state-supervised training. The inverse-dynamics model is only used during training, and SWIM improves planning success over uniform state weighting across five simulated robotic manipulation tasks.

JEPAAI summary
2026-10-28 · arXiv:2610.37391

JEPA-Inspired Latent Field Prediction for Sparse-Pilot MIMO-OFDM Channel Reconstruction

Yuzhi Yang, Brahim Mefgouda, Hang Zou et al.

The authors propose a JEPA-inspired field world model (FWM) for sparse-pilot MIMO-OFDM channel reconstruction. CSI is tokenized into latent time-frequency tokens; the model predicts masked/future latent field states from historical multi-band context, then fuses the predicted latent prior with sparse noisy pilots to reconstruct full CSI. Ray-tracing experiments show improved NMSE over pilot-only and direct reconstruction baselines.

JEPAAI summary