Timeline for Keras BatchNormalization axis
Current License: CC BY-SA 4.0
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 10, 2020 at 15:48 | comment | added | n1k31t4 |
It sounds reasonable to normalise over the frequency as it is your feature of interest (that means BatchNormalization takes the mean/variance over time and your "colour" channel. Again, I think time alignment of the samples might be important. However, I honestly am not sure what could make most sense with your data as it will likely depend on what you are trying to predict. I would suggest trying all three possibilities (it is only changing one value and re-running the experiment) and assess the impact. Maybe you will be able to then explain the differences fully to us all :)
|
|
| Feb 10, 2020 at 15:36 | comment | added | E. Vasilopoulos | So if i use axis=1 corresponding to the "time" dimension is more reasonable than what I suggested earlier, correct? | |
| Feb 10, 2020 at 14:46 | vote | accept | E. Vasilopoulos | ||
| Feb 10, 2020 at 14:38 | history | answered | n1k31t4 | CC BY-SA 4.0 |