Diffusion Operator Geometry of Feedforward Representations
Abstract
Feedforward neural networks transform data through learned representations whose geometry shapes how classes separate and relate across successive layers. We study that geometry through diffusion operators. Each feature-cloud snapshot is assigned a Gaussian-kernel Markov operator, giving a smooth description of one-step transport between classes from which spectral, boundary, and local geometric information can be read. We define the empirical class chain, state the condition under which it is an exact Markov quotient, and derive both the corresponding population transition and a simpler overlap chain based on expected class affinities. For balanced shared-covariance Gaussian class-conditional snapshots these affinities have closed forms controlled by a regularized Mahalanobis separation, which yields explicit expressions for leakage and coarse spectral behaviour. We further show that operator observables vary smoothly under feature perturbations, whereas hard neighborhood graphs are controlled by neighbor-order margins. Experiments on CIFAR-10 and CIFAR-100 ResNet-18 representations find that class transport becomes increasingly persistent with depth while retaining structured relations between classes, and that the diffusion class chain is more stable than its k-nearest-neighbor counterpart under matched perturbations.
Related
- Operator representations of Herglotz-Nevanlinna functions in several variables
- Same stimuli, different spaces: how task, observer and amount of data influence geometry and reliability of mental representations
- fSRD: Fuzzy Spectral Region Decomposition -- Automated Multi Operator Koopman Representations via an Adaptive Spectral Learning Architecture
- Diffusion-Based Stochastic Operator Networks for Uncertainty Quantification in Stochastic Partial Differential Equations
- Translating Between the Representations of an Acyclic Convex Geometry of Bounded Degree
BibTeX
@misc{reddy2026diffusion,
title = {{Diffusion Operator Geometry of Feedforward Representations}},
author = {Kanishka Reddy},
howpublished = {arXiv},
year = {2026},
month = aug,
doi = {10.48550/arxiv.2605.01107},
eprint = {2605.01107},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2605.01107},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Diffusion Operator Geometry of Feedforward Representations
AU - Kanishka Reddy
PY - 2026
DA - 2026/08/06
PB - arXiv
DO - 10.48550/arxiv.2605.01107
UR - https://arxiv.org/abs/2605.01107
AB - Feedforward neural networks transform data through learned representations whose geometry shapes how classes separate and relate across successive layers. We study that geometry through diffusion operators. Each feature-cloud snapshot is assigned a Gaussian-kernel Markov operator, giving a smooth description of one-step transport between classes from which spectral, boundary, and local geometric information can be read. We define the empirical class chain, state the condition under which it is an exact Markov quotient, and derive both the corresponding population transition and a simpler overlap chain based on expected class affinities. For balanced shared-covariance Gaussian class-conditional snapshots these affinities have closed forms controlled by a regularized Mahalanobis separation, which yields explicit expressions for leakage and coarse spectral behaviour. We further show that operator observables vary smoothly under feature perturbations, whereas hard neighborhood graphs are controlled by neighbor-order margins. Experiments on CIFAR-10 and CIFAR-100 ResNet-18 representations find that class transport becomes increasingly persistent with depth while retaining structured relations between classes, and that the diffusion class chain is more stable than its k-nearest-neighbor counterpart under matched perturbations.
ER -