Timeline for Does it make sense to use `logit` or `softplus` loss for binary classification problem?
Current License: CC BY-SA 4.0
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 26, 2023 at 2:25 | comment | added | THN | @K_Augus They are equivalent when L1 is the cross-entropy loss for binary classification. It can be shown by splitting the positive and negative data. You can read the appendix of this paper (arxiv.org/abs/1903.11406) for a reference. | |
| Apr 23, 2023 at 19:08 | comment | added | K_Augus | Could you explain a little more about why L3 is equivalient to L1 | |
| Nov 8, 2018 at 8:13 | vote | accept | THN | ||
| Nov 8, 2018 at 5:50 | history | answered | THN | CC BY-SA 4.0 |