data_mining:neural_network:initialization

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
data_mining:neural_network:initialization [2017/08/19 22:37] – [Random initialization] phreazerdata_mining:neural_network:initialization [2017/08/19 22:38] (current) – [Random initialization] phreazer
Line 8: Line 8:
 Solution: $W^{[i]}=np.random.randn((2,2)) * 0.01$ Solution: $W^{[i]}=np.random.randn((2,2)) * 0.01$
  
-$0.01$ because else we would end up at ends of activation function values (and slopes would be small).+$0.01$ because else we would end up at ends of activation function values (and slopes would be small), e.g. if values would be large.
  • data_mining/neural_network/initialization.1503175053.txt.gz
  • Last modified: 2017/08/19 22:37
  • by phreazer