Timeline for Neural Net Backprop Weight updating Pseudo code help please
Current License: CC BY-SA 4.0
2 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| May 28, 2020 at 11:32 | comment | added | indigo | Hi, thankyou for your reply. I have the same number of inputs as hidden nodes and just 1 hidden layer and 1 output node. I added the output code above. I find now the error goes down slightly in the order of 0.0001 in the first handful of iterations but then stops going down much at all and with higher learning rates and initial weights actually goes up with very small output values nowhere near the target. Will bias units help this much? | |
| May 27, 2020 at 22:11 | history | answered | Shan S | CC BY-SA 4.0 |