loading page

Geometric Back-Propagation in Morphological Neural Networks
  • Rick Groenendijk ,
  • L. Dorst ,
  • T. Gevers
Rick Groenendijk
University of Amsterdam

Corresponding Author:[email protected]

Author Profile
T. Gevers
Author Profile


This paper provides a definition of back-propagation through geometric correspondences for morphological neural networks. In addition, dilation layers are shown to learn probe geometry by erosion of layer inputs and outputs. A proof-of-principle is provided, in which predictions and convergence of morphological networks significantly outperform convolutional networks.
2023Published in IEEE Transactions on Pattern Analysis and Machine Intelligence on pages 1-8. 10.1109/TPAMI.2023.3290615