logo

SCIENCE CHINA Information Sciences, Volume 64 , Issue 2 : 129101(2021) https://doi.org/10.1007/s11432-018-1513-5

RGA-CNNs: convolutional neural networks based on reduced geometric algebra

More info
  • ReceivedDec 10, 2018
  • AcceptedJun 22, 2019
  • PublishedJul 16, 2020

Abstract

There is no abstract available for this article.


Acknowledgment

This work was supported by National Natural Science Foundation of China (Grant Nos. 61771299, 61771322, 61375015, 61301027)


Supplement

Appendixes A–C.


References

[1] Girshick R, Donahue J, Darrell T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2014. 580--587. Google Scholar

[2] Taigman Y, Yang M, Ranzato M, et al. DeepFace: closing the gap to human-level performance in face verification. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2014. 1701--1708. Google Scholar

[3] Krizhevsky A, Sutskever L, Hinton G E. Imagenet classification with deep convolutional neural networks. In: Proceedings of Advances in Neural Information Processing Systems, 2012. 1097--1105. Google Scholar

[4] Zhang F, Cai N, Wu J X. Image denoising method based on a deep convolution neural network. Iet Image Process, 2018, 12: 485-493 CrossRef Google Scholar

[5] Guberman N. On Complex Valued Convolutional Neural Networks. 2016,. arXiv Google Scholar

[6] Kominami Y, Ogawa H, Murase K. Convolutional neural networks with multi-valued neurons. In: Proceedings of International Joint Conference on Neural Networks, Anchorage, 2017. 2673--2678. Google Scholar

[7] Shen M M, Wang R, Cao W M. Joint Sparse Representation Model for Multi-Channel Image Based on Reduced Geometric Algebra. IEEE Access, 2018, 6: 24213-24223 CrossRef Google Scholar

[8] Graham B. Fractional Max-Pooling. 2015,. arXiv Google Scholar

[9] Springenberg J T, Dosovitskiy A, Brox T, et al. Striving for simplicity: the all convolutional net. In: Proceedings of International Conference on Learning Representations (ICLR), 2015. 1--14. Google Scholar

  • Figure 1

    (Color online) Training loss curves achieved by the real-valued CNN, QCNN, and proposed RGA-CNN forprotectłinebreak (a) the 3D geometrical shape and (b) the color image classification tasks.