mlsanityData to 7 Oct 2026

SJEPA: Learning Elegant Latent Dynamics with Hybrid Symbolic-Neural Predictors

arXiv 2608.04060 ↗First posted 4 Aug 2026cs.LG cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Joint-embedding predictive architectures learn abstract states by predicting target embeddings from context embeddings, but their transition models are typically opaque neural maps. We introduce SJEPA, a reconstruction-free JEPA framework that learns predictive representations whose induced dynamics admit compact symbolic descriptions. Its hybrid transition combines a symbolic law with a regularised neural correction for dynamics outside the selected grammar. The central principle is to learn the simplest adequate dynamics: representation constraints preserve informative, non-collapsed predictive coordinates, while operator compression favours low-complexity symbolic-neural transitions that remain predictively adequate. We formalise this principle through induced-dynamics complexity, analyse predictive-coordinate non-identifiability, and show that unconstrained operator compression creates a direct shortcut to representation collapse. The framework supports both alternating representation-equation learning and symbolic dynamics fitted to fixed representations. In controlled pendulum experiments, joint learning discovers substantially simpler symbolic dynamics with lower long-horizon rollout error and divergence than post-hoc fitting, while an unconstrained one-step diagnostic realises the predicted collapse shortcut. Under grammar misspecification, correction regularisation preserves the representable symbolic mechanism and directs the neural component towards residual dynamics. The results expose a controllable trade-off among predictive fidelity, representation quality, symbolic parsimony, and symbolic-neural allocation.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 26 of its 37 references that are in the map. First posted on arXiv 4 Aug 2026.

Builds on

26 of 37 references.

References · 26

  1. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  2. 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,339
  3. 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,893
  4. 2020Bootstrap your own latent: A new approach to self-supervised Learningcited by 4,712
  5. 2018Neural Ordinary Differential Equationscited by 3,733
  6. 2021Barlow Twins: Self-Supervised Learning via Redundancy Reductioncited by 1,818
  7. 2021VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learningcited by 1,013
  8. 2023Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecturecited by 982
  9. 2016Deep Variational Information Bottleneckcited by 915
  10. 2018Recurrent World Models Facilitate Policy Evolutioncited by 761
  11. 2017Deep learning for universal linear embeddings of nonlinear dynamicscited by 491
  12. 2024Revisiting Feature Prediction for Learning Visual Representations from Videocited by 490
14 more
  1. 2020Discovering Symbolic Models from Deep Learning with Inductive Biasescited by 335
  2. 2019Deep symbolic regression: Recovering mathematical expressions from data via risk-seeking policy gradientscited by 294
  3. 2025LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristicscited by 227
  4. 2021Neural Symbolic Regression that Scalescited by 185
  5. 2017Sparse identification of nonlinear dynamics for model predictive control in the low-data limitcited by 139
  6. 2023MC-JEPA: A Joint-Embedding Predictive Architecture for Self-Supervised Learning of Motion and Content Featurescited by 53
  7. 2025Audio-JEPA: Joint-Embedding Predictive Architecture for Audio Representation Learningcited by 17
  8. 20243D-JEPA: A Joint Embedding Predictive Architecture for 3D Self-Supervised Representation Learningcited by 14
  9. 2026VJEPA: Variational Joint Embedding Predictive Architectures as Probabilistic World Modelscited by 14
  10. 2025ACT-JEPA: Novel Joint-Embedding Predictive Architecture for Efficient Policy Representation Learningcited by 7
  11. 2025T-SHRED: Symbolic Regression for Regularization and Model Discovery with Transformer Shallow Recurrent Decoderscited by 4
  12. 2026From Observation to Insight: Mechanistic World Models and the Quest for Autonomous Discoverycited by 4
  13. 2026Knowledge, Rules and Their Embeddings: Two Paths towards Neuro-Symbolic JEPAcited by 1
  14. 2026Extracting Governing Equations from Latent Dynamics via Multi-View Contrastive Learningcited by 1

Built on by

Not cited yet.