Skip to main content

Table 3 Reconstruction SDR with/without regularization in decibel

From: Discriminative frequency filter banks learning with neural networks

Method

M = 32, R = T

M = 64, R = T

M = 64, R = F

TriFB

8.45

13.01

12.92

GaussFB

8.12

12.44

12.44

TriFB-DN

10.32

14.69

13.28

GaussFB-DN

9.55

12.92

12.22

BandPosFB-DN

8.57

12.84

12.15

PosFB-DN

14.44

18.24

17.21

  1. M represents the number of filters, R represents the regularization option, T represents true, and F represents false