Rotation Equivariance and Invariance in Convolutional Neural Networks

Abstract

Performance of neural networks can be significantly improved by encoding known invariance for particular tasks. Many image classification tasks, such as those related to cellular imaging, exhibit invariance to rotation. We present a novel scheme using the magnitude response of the 2D-discrete-Fourier transform (2D-DFT) to encode rotational invariance in neural networks, along with a new, efficient convolutional scheme for encoding rotational equivariance throughout convolutional layers. We implemented this scheme for several image classification tasks and demonstrated improved performance, in terms of classification accuracy, time required to train the model, and robustness to hyperparameter selection, over a standard CNN and another state-of-the-art method.

References in corpus (1)

• Deformable Convolutional Networks

Cited by in corpus (7)

• Incorporating Symmetry into Deep Dynamics Models for Improved Generalization
• Trajectory Prediction using Equivariant Continuous Convolution
• Deep Learning for 2D and 3D Rotatable Data: An Overview of Methods
• A simple equivariant machine learning method for dynamics based on scalars
• Geared Rotationally Identical and Invariant Convolutional Neural Network Systems
• Transformationally Identical and Invariant Convolutional Neural Networks through Symmetric Element Operators
• Self-Refining Deep Symmetry Enhanced Network for Rain Removal