mlsanityData to 14 Oct 2026

Beyond Local Linearity: Scale-Resolved Geometry of Learned Image Encoders

arXiv 2609.39115 ↗First posted 30 Sep 2026cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Understanding how learned representations respond to finite input changes is important for characterizing their sensitivity, invariances, and robustness. Yet existing geometric analyses are predominantly local and describe only infinitesimal perturbations. We introduce a scale-resolved statistic that compares an encoder's measured feature displacement with its local linear prediction as the perturbation magnitude increases. Across diverse image encoders, we discover a characteristic plateau-rise-peak-decay profile, which we call the bump. The bump is absent at initialization, emerges early during standard training, and does not form under randomized labels or random-noise inputs. Its shape also varies with the training distribution and robustness objective. These results establish departures from local geometry as a signature of how encoder representations are shaped by learning.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 23 of its 27 references that are in the map. First posted on arXiv 30 Sep 2026.

Builds on

23 of 27 references.

References · 23

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  3. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,863
  4. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
  5. 2014ImageNet Large Scale Visual Recognition Challengecited by 8,498
  6. 2018Neural Tangent Kernel: Convergence and Generalization in Neural Networkscited by 2,621
  7. 2019Natural Adversarial Examplescited by 1,408
  8. 2018On Lazy Training in Differentiable Programmingcited by 675
  9. 2020Do Adversarially Robust ImageNet Models Transfer Better?cited by 316
  10. 2016On the Expressive Power of Deep Neural Networkscited by 306
  11. 2016Exponential expressivity in deep neural networks through transient chaoscited by 261
  12. 2018Sensitivity and Generalization in Neural Networks: an Empirical Studycited by 189
11 more
  1. 2017Latent Space Oddity: on the Curvature of Deep Generative Modelscited by 174
  2. 2022Fake it till you make it: Learning transferable representations from synthetic ImageNet clonescited by 172
  3. 2019Complexity of Linear Regions in Deep Networkscited by 167
  4. 2019Deep ReLU Networks Have Surprisingly Few Activation Patternscited by 161
  5. 2019Adversarial Robustness through Local Linearizationcited by 146
  6. 2017The Riemannian Geometry of Deep Generative Modelscited by 82
  7. 2017Metrics for Deep Generative Modelscited by 53
  8. 2024Deep Networks Always Grok and Here is Whycited by 51
  9. 2014Metrics for Probabilistic Geometriescited by 28
  10. 2017Eigen-Distortions of Hierarchical Representationscited by 26
  11. 2024Discriminating image representations with principal distortionscited by 8

Built on by

Not cited yet.