Spectral characteristics of autoencoder parameters as a vector representation of data
Abstract
This paper examines the relationship between the parameters of autoencoder models and the statistical properties of the data on which they are trained. Autoencoders are defined as models with an encoder-decoder architecture, trained to reconstruct input data through a compressed latent representation. It is proposed that the model parameters can be viewed as a dense vector representation of the corresponding sample. To test this hypothesis, a theoretical and experimental study is conducted in which a vector representation is formed based on the spectral characteristics of the autoencoder parameter matrices. Theoretical analysis shows that the singular values of the model parameter matrices are related to the eigenvalues of the covariance matrix of the training data, ensuring the transfer of information between the data space and the parameter space. Experimental results on the CIFAR-10 and FashionMNIST datasets confirm that the resulting vector representations allow for a high degree of accuracy in distinguishing between models trained on different data subsets, without resorting to complex vector generation algorithms or using the original samples. These results suggest that the parameters of trained autoencoders can be viewed as sample representations.
Versions
Related
- Topological Characteristics for the Analysis of Vector Fields. New stable characteristics for discrete and continuous dynamical systems
- Extremal Graphs with Prescribed Parameters for the Spectral Radius of Weighted Adjacency Matrices with Property P∗
- Robust acoustic representation of Japanese vowels under phonation variability using dynamic spectral analysis
- Spectral Structures on Vector Bundles
- Sharp Regret Bounds and a Task-Covariance Correction for Spectral Representation Learning
BibTeX
@misc{nikitina2026spectral,
title = {{Spectral characteristics of autoencoder parameters as a vector representation of data}},
author = {Maria Nikitina and Anton Bishuk and Oleg Bakhteev},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.03495},
eprint = {2609.03495},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.03495},
note = {CC BY-NC-SA 4.0},
}
RIS
TY - UNPB
TI - Spectral characteristics of autoencoder parameters as a vector representation of data
AU - Maria Nikitina
AU - Anton Bishuk
AU - Oleg Bakhteev
PY - 2026
DA - 2026/09/03
PB - arXiv
DO - 10.48550/arxiv.2609.03495
UR - https://arxiv.org/abs/2609.03495
AB - This paper examines the relationship between the parameters of autoencoder models and the statistical properties of the data on which they are trained. Autoencoders are defined as models with an encoder-decoder architecture, trained to reconstruct input data through a compressed latent representation. It is proposed that the model parameters can be viewed as a dense vector representation of the corresponding sample. To test this hypothesis, a theoretical and experimental study is conducted in which a vector representation is formed based on the spectral characteristics of the autoencoder parameter matrices. Theoretical analysis shows that the singular values of the model parameter matrices are related to the eigenvalues of the covariance matrix of the training data, ensuring the transfer of information between the data space and the parameter space. Experimental results on the CIFAR-10 and FashionMNIST datasets confirm that the resulting vector representations allow for a high degree of accuracy in distinguishing between models trained on different data subsets, without resorting to complex vector generation algorithms or using the original samples. These results suggest that the parameters of trained autoencoders can be viewed as sample representations.
ER -