Timeline for How to intuitively explain what a kernel is?
Current License: CC BY-SA 4.0
13 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 24, 2021 at 9:14 | comment | added | Daniel Wiczew | @XavierBourretSicotte So in general, when I'd project my data by a kernel, I would project them from a space of data, to space of similarities between the data ? But then for 100 datapoints, number of similarities would be $\frac{100!}{2!(100-2)!}$ or a 100x100 matrix. | |
| Jul 23, 2021 at 19:55 | comment | added | Xavier Bourret Sicotte | @DanielWiczew x dash is a different vector, I could have called it y for example | |
| Jul 23, 2021 at 19:54 | comment | added | Xavier Bourret Sicotte | @Oliver yes they are called Mercer's theorem, see here for a summary: xavierbourretsicotte.github.io/Kernel_feature_map.html | |
| Jul 23, 2021 at 14:40 | comment | added | Daniel Wiczew | what is $x'$, it's another datapoint in the dataset ? We chose it arbitraly ? Or we compare each point in the dataset with another ? | |
| Jul 15, 2021 at 16:13 | comment | added | Oliver | Are there general rules on how to show that a function is a valid kernel? | |
| Apr 16, 2020 at 12:51 | comment | added | PascalIv | Sure it will work with the mapping z = x^2 +y^2 and since this is even lower-dimensional than the original data, using the kernel view would not make any sense. Therefore your example does not really motivate the use of kernels. But I get that usually, we do not see that easily, which nonlinear mapping to use. | |
| Apr 4, 2020 at 14:58 | comment | added | PascalIv | "f we could find a higher dimensional space in which these points were linearly separable" Why not take a lower-dimensional space e.g. the radius of the points? | |
| S Jan 27, 2020 at 19:32 | history | suggested | Barblog | CC BY-SA 4.0 |
corrected one typo x1 was written as x2, and added commas in the row vector to clearly separate the 3 elements
|
| Jan 27, 2020 at 17:56 | review | Suggested edits | |||
| S Jan 27, 2020 at 19:32 | |||||
| S Oct 17, 2018 at 19:13 | history | suggested | Robert Lugg | CC BY-SA 4.0 |
There is a bug in StackExchange when mathjax is placed in a quote (ie '>'). It places the > in the actual formula! I fixed that here by just putting everything on one line.
|
| Oct 17, 2018 at 17:52 | review | Suggested edits | |||
| S Oct 17, 2018 at 19:13 | |||||
| Aug 30, 2018 at 12:38 | history | edited | Xavier Bourret Sicotte | CC BY-SA 4.0 |
added link to code
|
| Jul 8, 2018 at 10:21 | history | answered | Xavier Bourret Sicotte | CC BY-SA 4.0 |