Theory of Deep Learning IIb: Optimization Properties of SGD
In Theory IIb we characterize with a mix of theory and experiments the optimization of deep convolutional networks by Stochastic Gradient Descent. The main new result in this paper is theoretical and experimental evidence for the following conjecture about SGD: SGD concentrates in probability - like...
| Main Authors: | , , , , , |
|---|---|
| Format: | Technical Report |
| Language: | en_US |
| Published: |
Center for Brains, Minds and Machines (CBMM)
2018
|
| Online Access: | http://hdl.handle.net/1721.1/115407 |