Timeline for Cost function of neural network is non-convex?
Current License: CC BY-SA 3.0
7 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 17, 2021 at 7:03 | comment | added | AlwaysLearning | Thank you! It's probably even better to cite the version for publication: proceedings.mlr.press/v38/choromanska15.pdf | |
| Feb 16, 2021 at 22:47 | comment | added | user20160 | @AlwaysLearning It's not always good, but the probability that it will be can increase with network size. Please have a look at the paper I mentioned. | |
| Feb 16, 2021 at 12:07 | comment | added | AlwaysLearning | My question is: how come this local minimum is always good in terms of fitting the training data. (given a reasonable depth and size of the network) | |
| Feb 16, 2021 at 2:36 | comment | added | user20160 | @AlwaysLearning Gradient-based optimization algorithms can indeed get stuck in local minima. The point is that this is ok if the local minimum you get stuck in is a good one which, in machine learning tasks, means the corresponding parameters give good generalization performance. | |
| Feb 15, 2021 at 19:52 | comment | added | AlwaysLearning | Then why does backpropagation work and not get stuck in a local minimum? | |
| Mar 21, 2020 at 16:44 | comment | added | Seymour | thank you for the reference | |
| May 23, 2016 at 8:21 | history | answered | user20160 | CC BY-SA 3.0 |