Timeline for Can we derive cross entropy formula as maximum likelihood estimation for SOFT LABELS?
Current License: CC BY-SA 4.0
6 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 13, 2020 at 12:14 | history | edited | Firebug | CC BY-SA 4.0 |
added 161 characters in body
|
| Oct 13, 2020 at 12:06 | comment | added | Firebug | @user20160 Your logical is right, except that you conclusion "grayscale is not a probability, thus Continuous Bernoulli does not apply to soft labels" is frail. If you treat $k$ as continuous, as they did, you need to include the correction factor, there is no escape from that. | |
| Oct 11, 2020 at 4:39 | comment | added | user20160 | But, I'd argue this means that such values are not soft labels at all. I consider soft labels to be actual probabilities given for different values a discrete variable may take. These can arise, for example, when training one probabilistic model to approximate another. Minimizing the cross entropy is indeed an appropriate thing to do in this situation. | |
| Oct 11, 2020 at 4:39 | comment | added | user20160 | The paper points out that people sometimes erroneously treat values in $[0,1]$ as probabilities defining a Bernoulli distribution. I wholeheartedly agree with this--grayscale values are not probabilities of an image pixel being 0 or 1. They're continuous observed values and should be modeled as such. (continued...) | |
| Oct 10, 2020 at 23:39 | history | edited | Firebug | CC BY-SA 4.0 |
deleted 6 characters in body
|
| Oct 10, 2020 at 22:50 | history | answered | Firebug | CC BY-SA 4.0 |