Input-to-hidden layers synapses are too large

I159 / go_deep

Neural network framework on pure Go

MIT License

1 stars 0 forks source link

Closed I159 closed 6 years ago

I159 commented 6 years ago

With Nguyen-Widrow algorithm adapted to input-to-hidden synapses layer (possibly wrong) produces very large numbers.