From CNNs to Shift-Invariant Twin Models Based on Complex Wavelets - Archive ouverte HAL Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2023

From CNNs to Shift-Invariant Twin Models Based on Complex Wavelets

Résumé

We propose a novel antialiasing method to increase shift invariance and prediction accuracy in convolutional neural networks. Specifically, we replace the first-layer combination "real-valued convolutions + max pooling" ($\mathbb{R}$Max) by "complex-valued convolutions + modulus'' ($\mathbb{C}$Mod), which is stable to translations. To justify our approach, we claim that $\mathbb{C}$Mod and $\mathbb{R}$Max produce comparable outputs when the convolution kernel is band-pass and oriented (Gabor-like filter). In this context, $\mathbb{C}$Mod can be considered as a stable alternative to $\mathbb{R}$Max. Thus, prior to antialiasing, we force the convolution kernels to adopt such a Gabor-like structure. The corresponding architecture is called mathematical twin, because it employs a well-defined mathematical operator to mimic the behavior of the original, freely-trained model. Our antialiasing approach achieves superior accuracy on ImageNet and CIFAR-10 classification tasks, compared to prior methods based on low-pass filtering. Arguably, our approach's emphasis on retaining high-frequency details contributes to a better balance between shift invariance and information preservation, resulting in improved performance. Furthermore, it has a lower computational cost and memory footprint than concurrent work, making it a promising solution for practical implementation.
Fichier principal
Vignette du fichier
preprint.pdf (2.52 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03880520 , version 1 (01-12-2022)
hal-03880520 , version 2 (21-04-2023)

Identifiants

  • HAL Id : hal-03880520 , version 2

Citer

Hubert Leterme, Kévin Polisano, Valérie Perrier, Karteek Alahari. From CNNs to Shift-Invariant Twin Models Based on Complex Wavelets. 2023. ⟨hal-03880520v2⟩
96 Consultations
38 Téléchargements

Partager

Gmail Facebook X LinkedIn More