Network Pruning on CNN
31 views (last 30 days)
More Answers (1)
Katja Mogalle on 18 May 2022
There is a new pruning feature in R2022a that let's you prune whole filters in 2D convolution layers. By removing unimportant filters in the network, the memory footprint of the network reduces and inference get's faster.
See taylorPrunableNetwork doc page for more information. There are also two examples:
- Prune Image Classification Network Using Taylor Scores
- Prune Filters in a Detection Network Using Taylor Scores
Hope this helps.