mlsanityData to 14 Oct 2026

KerJEPA: Kernel Discrepancies for Euclidean Self-Supervised Learning

arXiv 2512.19605 ↗First posted 22 Dec 2025cs.LG cs.CVcited by 8

Not on the map. Why?

Abstract (arXiv)

Breakthroughs in self-supervised Joint-Embedding Predictive Architectures (JEPAs) have established that regularizing Euclidean representations toward isotropic Gaussian priors yields provable gains in training stability and downstream generalization. We introduce a new, flexible family of Euclidean joint embedding self-supervised learning algorithms with kernel-based regularizers. One instance of this family corresponds to the SIGReg Epps-Pulley regularizer which approximates a sliced Maximum Mean Discrepancy (MMD) with a Gaussian prior and Gaussian kernel. By expanding the class of viable kernels and priors and computing the closed-form high-dimensional limit of sliced discrepancies, we develop alternative methods with improved training stability and design flexibility and leverage those techniques to train strong vision learners at scale.

Key facts. Cited by 8 papers in the map (data through 14 Oct 2026). Builds on 25 of its 38 references that are in the map. First posted on arXiv 22 Dec 2025.

Builds on

25 of 38 references.

References · 25

  1. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 12,049
  2. 2019Momentum Contrast for Unsupervised Visual Representation Learningcited by 7,513
  3. 2020Bootstrap your own latent: A new approach to self-supervised Learningcited by 4,750
  4. 2020Unsupervised Learning of Visual Features by Contrasting Cluster Assignmentscited by 2,857
  5. 2021Barlow Twins: Self-Supervised Learning via Redundancy Reductioncited by 1,838
  6. 2020Understanding Contrastive Representation Learning through Alignment and Uniformity on the Hyperspherecited by 1,405
  7. 2021VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learningcited by 1,057
  8. 2025LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristicscited by 254
  9. 2016A Kernelized Stein Discrepancy for Goodness-of-fit Tests and Model Evaluationcited by 236
  10. 2016A Kernel Test of Goodness of Fitcited by 143
  11. 2017Measuring Sample Quality with Kernelscited by 119
  12. 2017The Cramer Distance as a Solution to Biased Wasserstein Gradientscited by 119
13 more
  1. 2015Measuring Sample Quality with Stein's Methodcited by 94
  2. 2020Statistical and Topological Properties of Sliced Probability Divergencescited by 85
  3. 2021Self-Supervised Learning with Kernel Dependence Maximizationcited by 77
  4. 2021Kernel Stein Discrepancy Descentcited by 57
  5. 2022Spherical Sliced-Wassersteincited by 31
  6. 2022Guillotine Regularization: Why removing layers is needed to improve generalization in Self-Supervised Learningcited by 26
  7. 2018Random Feature Stein Discrepanciescited by 23
  8. 2020Sliced Kernelized Stein Discrepancycited by 23
  9. 2014FastMMD: Ensemble of Circular Discrepancy for Efficient Two-Sample Testcited by 17
  10. 2024You Don't Need Domain-Specific Data Augmentations When Scaling Self-Supervised Learningcited by 16
  11. 2022Targeted Separation and Convergence with Kernel Discrepanciescited by 15
  12. 2023Compress Then Test: Powerful Kernel Testing in Near-linear Timecited by 14
  13. 2024Debiased Distribution Compressioncited by 8

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 8.

8 papers.

  1. 2026Curriculum-Driven 3D CT Report Generation via Language-Free Visual Grafting and Zone-Constrained Compressioncited by 3
  2. 2026Beyond Isotropy in JEPAs: Hamiltonian Geometry and Symplectic Predictioncited by 3
  3. 2026Minimax Lower Bounds of Kernel Discrepancy Estimation: MMD, HSIC, KSDcited by 2
  4. 2026Beyond Gaussian Worlds: Latent Geometry Matters for JEPAscited by 1
  5. 2026Adaptive Latent Capacity for World Modelscited by 0
  6. 2026Anisotropic Representations Improve Planning in JEPA World Modelscited by 0
  7. 2026Shared Gaussianization: What Gaussian Regularizers Certify About Contrastive Learning, and What They Misscited by 0
  8. 2026Kernel-based External Validity of Experimental Studiescited by 0