Lattice theory and algebraic models for deep convolutional learning based on mathematical morphology

Gustavo (Jesús) Angulo · 2026

Read Lattice theory and algebraic models for deep convolutional learning based on mathematical morphology by Gustavo (Jesús) Angulo (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-24608.txt.

Introduction

Gustavo Angulo develops a lattice-theoretic framework for deep convolutional architectures, including CNNs, ResNets, and encoder-decoder nets like UNet. The main tool is MMBB representation theory for translation-invariant maps, applied layer by layer. The key finding recasts the standard pipeline as a cross-lattice device. Convolution acts as erosion, ReLU as a lattice-join closing, and max-pooling as dilation, composing into a morphological opening in neither lattice. Since the composed layer is not idempotent, depth adds genuine power. Three idempotent opening designs are identified and characterized.

Catalog record

CREATORGustavo (Jesús) Angulo
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE227913
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.24608

Browse the full library · About the Tower of Babel