Learning Steerable Filters for Rotation Equivariant CNNs
arXiv:1711.07289
Abstract
In many machine learning tasks it is desirable that a model's prediction transforms in an equivariant way under transformations of its input. Convolutional neural networks (CNNs) implement translational equivariance by construction; for other transformations, however, they are compelled to learn the proper mapping. In this work, we develop Steerable Filter CNNs (SFCNNs) which achieve joint equivariance under translations and rotations by design. The proposed architecture employs steerable filters to efficiently compute orientation dependent responses for many orientations without suffering interpolation artifacts from filter rotation. We utilize group convolutions which guarantee an equivariant mapping. In addition, we generalize He's weight initialization scheme to filters which are defined as a linear combination of a system of atomic filters. Numerical experiments show a substantial enhancement of the sample complexity with a growing number of sampled filter orientations and confirm that the network generalizes learned patterns over orientations. The proposed approach achieves state-of-the-art on the rotated MNIST benchmark and on the ISBI 2012 2D EM segmentation challenge.
Camera ready version, accepted for CVPR 2018
References in corpus (2)
Cited by in corpus (8)
- Fanaroff-Riley classification of radio galaxies using group-equivariant convolutional neural networks
- VV-Net: Voxel VAE Net with Group Convolutions for Point Cloud Segmentation
- Scale equivariance in CNNs with vector fields
- Robustness of Rotation-Equivariant Networks to Adversarial Perturbations
- RotDCF: Decomposition of Convolutional Filters for Rotation-Equivariant Deep Networks
- Rotational 3D Texture Classification Using Group Equivariant CNNs
- Equivariant Entity-Relationship Networks
- Transformationally Identical and Invariant Convolutional Neural Networks by Combining Symmetric Operations or Input Vectors