scholarly journals On the Structural Sensitivity of Deep Convolutional Networks to the Directions of Fourier Basis Functions

Author(s):  
Yusuke Tsuzuku ◽  
Issei Sato
2021 ◽  
Vol 25 (4) ◽  
Author(s):  
Jelena Đorđević Kozarov ◽  
Atila Juhas ◽  
Dragan Pejić ◽  
Platon Sovilj ◽  
Vladimir Vujičić

Paper defines an algorithm for generating two - bit dithered discrete Fourier basis functions (2BDDFBF) used in a Stochastic Digital Discret Fourier Transformation (SDDFT) processor.  Based on the theoretical criterion of marginal precision, and norm and orthogonality, the orthonormality of DFT with 32 harmonics was confirmed by simulation and experimentally. The experiment was detailed and comprehensive, both for standardization and for both types of orthogonality. It was performed in 236800000 points in each of the three variants of orthonormalization. The matching of theoretical and experimental precision is very acceptable and it can be said with great reliability that the proposed algorithm for generating DDFBF is correct. 2BDDFBF play key role in electricity measurement what is emphasized in the paper. .


Sign in / Sign up

Export Citation Format

Share Document