Timeline for In a neural network, is it possible to gradient descent with more than one input?
Current License: CC BY-SA 4.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| S Oct 1, 2021 at 12:06 | vote | accept | Gergő Horváth | ||
| Oct 1, 2021 at 12:06 | comment | added | Gergő Horváth | @N.Osil I uploaded the code as answer. | |
| Oct 1, 2021 at 12:05 | answer | added | Gergő Horváth | timeline score: 0 | |
| Sep 29, 2021 at 20:11 | comment | added | N. Osil | Funny how this question looks very basic but it is really not. You really tried to implement backpropagation, I hope you share the code. Short answer: you cannot. Long answer is that in libraries like Tensorflow, the gradients are computed for each example, not for the averaged loss. Then each point is averaged. I will try to make a notebook and add the link here to show how it really works. | |
| Aug 20, 2020 at 13:22 | vote | accept | Gergő Horváth | ||
| S Oct 1, 2021 at 12:06 | |||||
| Aug 20, 2020 at 6:26 | answer | added | Mayank Mishra | timeline score: 1 | |
| Jul 29, 2020 at 11:14 | answer | added | 10xAI | timeline score: 0 | |
| Jul 28, 2020 at 20:16 | history | asked | Gergő Horváth | CC BY-SA 4.0 |