Rotation Equivariance and Invariance in Convolutional Neural Networks
This work addresses rotation invariance in image classification, particularly for domains like cellular imaging, offering incremental improvements over existing methods.
The authors tackled the problem of improving neural network performance for rotation-invariant image classification tasks, such as cellular imaging, by introducing a novel scheme using the 2D-DFT magnitude for invariance and an efficient convolutional method for equivariance, resulting in enhanced classification accuracy, faster training times, and better robustness compared to standard and state-of-the-art methods.
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.