Quaternion convolutional neural networks for heterogeneous image processing

Abstract : Convolutional neural networks (CNN) have recently achieved state-of-the-art results in various applications. In the case of image recognition, an ideal model has to learn independently of the training data, both local dependencies between the three components (R,G,B) of a pixel, and the global relations describing edges or shapes, making it efficient with small or heterogeneous datasets. Quaternion-valued convo-lutional neural networks (QCNN) solved this problematic by introducing multidimensional algebra to CNN. This paper proposes to explore the fundamental reason of the success of QCNN over CNN, by investigating the impact of the Hamilton product on a color image reconstruction task performed from a gray-scale only training. By learning independently both internal and external relations and with less parameters than real valued convolutional encoder-decoder (CAE), quaternion convolutional encoder-decoders (QCAE) perfectly reconstructed unseen color images while CAE produced worst and gray-scale versions. Index Terms-Quaternion convolutional encoder-decoder, convolutional neural networks, heterogeneous image processing
Document type :
Conference papers
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-02107644
Contributor : Titouan Parcollet <>
Submitted on : Tuesday, April 23, 2019 - 5:34:38 PM
Last modification on : Friday, May 10, 2019 - 1:25:57 AM

File

QCNN_for_Color_Image-2.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02107644, version 1
  • ARXIV : 1811.02656

Collections

Citation

Titouan Parcollet, Mohamed Morchid, Georges Linarès. Quaternion convolutional neural networks for heterogeneous image processing. IEEE ICASSP, May 2019, Brighton, United Kingdom. ⟨hal-02107644⟩

Share

Metrics

Record views

4

Files downloads

7