Abstract:We present a PDE-based framework that generalizes Group equivariant Convolutional Neural Networks (G-CNNs). In this framework, a network layer is seen as a set of PDE-solvers where geometrically meaningful PDE-coefficients become the layer's trainable weights. Formulating our PDEs on homogeneous spaces allows these networks to be designed with built-in symmetries such as rotation in addition to the standard translation equivariance of CNNs. Having all the desired symmetries included in the design obviates the need to include them by means of costly techniques such as data augmentation. We will discuss our PDE-based G-CNNs (PDE-G-CNNs) in a general homogeneous space setting while also going into the specifics of our primary case of interest: roto-translation equivariance. We solve the PDE of interest by a combination of linear group convolutions and non-linear morphological group convolutions with analytic kernel approximations that we underpin with formal theorems. Our kernel approximations allow for fast GPU-implementation of the PDE-solvers, we release our implementation with this article in the form of the LieTorch extension to PyTorch, available at <a class="link-external link-https" href="https://gitlab.com/bsmetsjr/lietorch" rel="external noopener nofollow">this https URL</a> . Just like for linear convolution a morphological convolution is specified by a kernel that we train in our PDE-G-CNNs. In PDE-G-CNNs we do not use non-linearities such as max/min-pooling and ReLUs as they are already subsumed by morphological convolutions. We present a set of experiments to demonstrate the strength of the proposed PDE-G-CNNs in increasing the performance of deep learning based imaging applications with far fewer parameters than traditional CNNs.

Group Equivariant Convolutional Networks

A Dynamic Group Equivariant Convolutional Networks for Medical Image Analysis.

Characterization and Design of Generalized Convolutional Neural Network

A General Framework for Robust G-Invariance in G-Equivariant Networks

PDE-based Group Equivariant Convolutional Neural Networks

Intertwiners between Induced Representations (with Applications to the Theory of Equivariant Neural Networks)

Variational Partial Group Convolutions for Input-Aware Partial Equivariance of Rotations and Color-Shifts

Efficient Equivariant Network

Generalizing Convolutional Neural Networks for Equivariance to Lie Groups on Arbitrary Continuous Data

Gauge Equivariant Convolutional Networks and the Icosahedral CNN

Enabling equivariance for arbitrary Lie groups

Transformation-invariant Gabor Convolutional Networks

Learning Color Equivariant Representations

B-Spline CNNs on Lie Groups

Group Action Equivariance And Generalized Convolution In Multi-Layer Neural Networks

Lie Group Convolution Neural Networks with Scale-Rotation Equivariance

Group Equivariant Subsampling

Gaussian Transfer Convolutional Neural Networks

Gradient-Aligned convolution neural network

Adaptive aggregation of Monte Carlo augmented decomposed filters for efficient group-equivariant convolutional neural network

Dynamic Group Convolution for Accelerating Convolutional Neural Networks