2 Hits in 3.8 sec

Grassmannian Packings in Neural Networks: Learning with Maximal Subspace Packings for Diversity and Anti-Sparsity [article]

Dian Ang Yap, Nicholas Roberts, Vinay Uday Prabhu
2019 arXiv   pre-print
Drawing inspiration from information theory and wireless communications, we demonstrate the intersection of coding theory and deep learning through the Grassmannian subspace packing problem in CNNs.  ...  We show that Grassmannian packings, especially in the initial layers, address kernel sparsity and encourage diversity, while improving classification accuracy across shallow and deep CNNs with better convergence  ...  Acknowledgments We would like to acknowledge Matthew McAteer for helpful discussions and contrbution of designs and plots for Figures 5 and 6 .  ... 
arXiv:1911.07418v1 fatcat:xufyag6rnfctjhbtgeqdphpdie

A Systematic Review on NOMA Variants for 5G and Beyond

Ishan Budhiraja, Neeraj Kumar, Sudhanshu Tyagi, Sudeep Tanwar, Zhu Han, Doug Young Suh, MD. Jalil Piran
2021 IEEE Access  
In this technique, spreading sequences are obtained by solving the Grassmannian line packing problem to maximize the minimum chordal distance between spreading codes.  ...  In contrast to [257] and [258] , the authors in [259] proposed a low complexity subspace detectors for THz MIMO-NOMA systems.  ... 
doi:10.1109/access.2021.3081601 fatcat:mzmmr3km5zehhh3fnupyelfo3e