A PDE-Based Convolutional Neural Network with Variational Information Bottleneck: Experimental Evaluation and Generalization Analysis
Mathematical Problems of Computer Science · 01.12.2025 · 64 · 37-46
DOI: 10.51408/1963-0138 ↗Kurz gefasst
Fachartikel in der Zeitschrift Mathematical Problems of Computer Science, online erschienen am 01.12.2025. Verfasst von Gor Gharagyozyan. Die Arbeit wurde bisher 1-mal zitiert (Crossref).
Abstract (Originalsprache)
We present a hybrid convolutional architecture that combines trainable PDE-based preprocessing with a Variational Information Bottleneck (VIB) to improve generalization in image classification. The PDE stage applies a small number of discretized Laplacian steps with learnable step size and depthwise coupling, injecting physics-inspired inductive bias into early feature maps. A tensor-wise VIB module then parameterizes a Gaussian latent (μ, log σ2) via 1×1 convolutions and enforces information compression through a KL penalty to a unit prior, encouraging retention of task-relevant features while discarding nuisance variability. The compressed representation feeds a ResNet-18 backbone adapted for CIFAR-10 inputs. On CIFAR-10, systematic variation of the VIB weight β shows that moderate compression yields improved test performance and training stability relative to both a baseline CNN and a PDE-only variant. Qualitative analysis indicates smoother activations and reduced sensitivity to input noise, consistent with the information-theoretic objective. The results suggest that PDE priors and variational compression act complementarily, offering a principled path to robust and generalizable convolutional models.
Zitieren
APA
Gharagyozyan, G. (2025). A PDE-Based Convolutional Neural Network with Variational Information Bottleneck: Experimental Evaluation and Generalization Analysis. Mathematical Problems of Computer Science, 64, 37-46. https://doi.org/10.51408/1963-0138
Vancouver
Gharagyozyan G. A PDE-Based Convolutional Neural Network with Variational Information Bottleneck: Experimental Evaluation and Generalization Analysis. Mathematical Problems of Computer Science. 2025;64:37-46. doi:10.51408/1963-0138
BibTeX
@article{Gharagyozyan2025,
author = {Gharagyozyan, Gor},
title = {A PDE-Based Convolutional Neural Network with Variational Information Bottleneck: Experimental Evaluation and Generalization Analysis},
journal = {Mathematical Problems of Computer Science},
year = {2025},
volume = {64},
pages = {37--46},
doi = {10.51408/1963-0138}
}
EUVIMED