Rotated Ring, Radial and Depth Wise Separable Radial Convolutions

Wolfgang Fuhl

Abstract:

Rotated Ring, Radial and Depth Wise Separable Radial Convolutions
Wolfgang Fuhl
University of T ̈ubingen
Sand 14, 72076 T ̈ubingen, Germany
[e-mail]
Enkelejda Kasneci
University of T ̈ubingen
Sand 14, 72076 T ̈ubingen, Germany
[e-mail]
Abstract
Simple image rotations significantly reduce the accu-
racy of deep neural networks. Moreover, training with
all possible rotations increases the data set, which also
increases the training duration. In this work, we address
trainable rotation invariant convolutions as well as the
construction of nets, since fully connected layers can only
be rotation invariant with a one-dimensional input. On
the one hand, we show that our approach is rotationally
invariant for different models and on different public data
sets. We also discuss the influence of purely rotational
invariant features on accuracy. The rotationally adaptive
convolution models presented in this work are more
computationally intensive than normal convolution models.
Therefore, we also present a depth wise separable approach
with radial convolution.