Rotation Invariance for Unsupervised Cell Representation Learning: Analysis of The Impact of Enforcing Rotation Invariance or Equivariance on Representation for Cell Classiﬁcation
Philipp Gräbel, Ina Laube, Martina Crysandt, Reinhild Herwartz, Melanie Baumann, Barbara M. Klinkhammer, Peter Boor, Tim H. Brümmendorf, Dorit Merhof
Rheinisch-Westfälische Technische Hochschule Aachen, Institute of Imaging & Computer Vision
While providing powerful solutions for many problems, deep neural networks require large amounts of training data. In medical image computing, this is a severe limitation, as the required expertise makes annotation efforts often infeasible. This also applies to the automated analysis of hematopoietic cells in bone marrow whole slide images. In this work, we propose approaches to restrict a neural network towards learning of rotation invariant or equivariant representation. Even though the proposed methods achieve this goal, it does not increase classification scores on unsupervisedly learned representations.
The author has not agreed to a public release of their video.