Skip to main content
. Author manuscript; available in PMC: 2019 Nov 27.
Published in final edited form as: Proc Mach Learn Res. 2019 Jun;97:1528–1537.

Figure 3:

Figure 3:

Accuracy gain relative to baseline (no augmentation) when averaging at various layers of a LeNet network. Approximation at earlier layers saves computation but can reduce the fidelity of the approximation.