Search Results for author:
Found 3 papers, 0 papers with code
Distribution Mismatch Correction for Improved Robustness in Deep Neural Networks
The most common normalization method, batch normalization, reduces the distribution shift during training but is agnostic to changes in the input distribution during test time.
Complex-valued Convolutional Neural Networks for Enhanced Radar Signal Denoising and Interference Mitigation
Our experiments show, that the use of CVCNNs increases data efficiency, speeds up network training and substantially improves the conservation of phase information during interference removal.
Wasserstein Routed Capsule Networks
Capsule networks offer interesting properties and provide an alternative to today's deep neural network architectures.
Cannot find the paper you are looking for? You can
Submit a new open access paper.
