Timeline for no decrease loss and val_loss
Current License: CC BY-SA 4.0
11 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Nov 14, 2020 at 10:38 | vote | accept | Paul | ||
| Nov 13, 2020 at 9:35 | history | migrated | from stackoverflow.com (revisions) | ||
| Nov 12, 2020 at 12:33 | comment | added | hafiz031 |
Use LeaklyReLU instead of ReLU and the problem will be fixed. Simply remove activation="relu" from Dense() and add another layer of LeaklyReLU after each of the Dense layers like: model.add(LeakyReLU(alpha=0.05)). I ran your code with this change for 100 times (n_repeat=100) and this problem didn't occur for a single time.
|
|
| Nov 12, 2020 at 9:06 | answer | added | Andre S. | timeline score: 3 | |
| Nov 11, 2020 at 20:14 | comment | added | Paul | @hafiz031 I have just update my github with 'univariate_test.ipynb', I use 'seed' for replicability and 'tensorflow. keras' as you, you can not that the third try doesn't learn anything. | |
| Nov 11, 2020 at 12:52 | comment | added | hafiz031 |
I have run your model and found no such problems in losses. The losses decrease perfectly. As @meTchaikovsky said in the answer, probably the problem is due to model initialization. Also I have used tensorflow.keras instead of keras
|
|
| Nov 11, 2020 at 10:43 | comment | added | Paul | I am agree to migrate the question, how to do it preserving the bounty? | |
| Nov 11, 2020 at 0:59 | comment | added | meTchaikovsky | @RomainReboulleau I agree, the problem is not caused by the code, I guess using a shallower model will resolve the problem. | |
| Nov 10, 2020 at 13:23 | answer | added | meTchaikovsky | timeline score: 3 | |
| Nov 9, 2020 at 19:54 | comment | added | Romain Reboulleau | You should ask this on Data Science Stack Exchange. I suggested the question to be migrated. | |
| Nov 5, 2020 at 22:30 | history | asked | Paul | CC BY-SA 4.0 |