An Adaptive Orthogonal Convolution Scheme for Efficient and Flexible CNN Architectures

Boissin, Thibaut, Mamalet, Franck, Fel, Thomas, Picard, Agustin Martin, Massena, Thomas, Serrurier, Mathieu

arXiv.org Artificial Intelligence 

Orthogonal convolutional layers are the workhorse of multiple areas in machine learning, such as adversarial robustness, normalizing flows, GANs, and Lipschitzconstrained models. Their ability to preserve norms and ensure stable gradient propagation makes them valuable for a large range of problems. Despite their promise, the deployment of orthogonal convolution in large-scale applications is a significant challenge due to computational overhead and limited support for modern features like strides, dilations, group convolutions, and transposed convolutions.In this paper, we introduce AOC (Adaptative Orthogonal Convolution), a scalable method for constructing orthogonal convolutions, effectively overcoming these limitations. This advancement unlocks the construction of architectures that were previously considered impractical. We demonstrate through our experiments that our method produces expressive models that become increasingly efficient as they scale. To foster further advancement, we provide an open-source library implementing this method, available at https://github.com/thib-s/orthogonium.

Duplicate Docs Excel Report

Title
None found

Similar Docs  Excel Report  more

TitleSimilaritySource
None found