Timeline for Grid search for SVM parameters; is this is really how it is done?
Current License: CC BY-SA 3.0
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 13, 2018 at 17:00 | comment | added | Hari | @DikranMarsupial Will warm start while selecting the bandwidth for the pre-image algorithm in kernel PCA? | |
| Nov 10, 2014 at 15:04 | comment | added | Dikran Marsupial | Gradient descent isn't actually that much faster than Nelder Mead simplex in practice as computing the gradient information takes time. There can also be a problem with over-fitting the model selection criterion, which means you don't want to optimise it too agressively, which is easy to do with gradient descent. | |
| Nov 10, 2014 at 14:01 | comment | added | bdeonovic | +1 for gradient descent, first thing I thought about when I read the question. | |
| Nov 10, 2014 at 12:54 | history | answered | Dikran Marsupial | CC BY-SA 3.0 |