Skip to main content
. 2017 Feb 8;13(2):e1005281. doi: 10.1371/journal.pcbi.1005281

Fig 8. Run-time comparison between AMA and AMA-SGD.

Fig 8

Minimization time as a function of training set size on: A linear-linear axes, and B log-log axes. Minimization time for AMA increases quadratically while AMA-SGD increases linearly (for a fixed batch size) with the number of elements in the training set. A comparison of AMA and AMA-SGD costs is shown in Fig 9.