Timeline for Understanding Gaussian Basis function parameters to be used in linear regression
Current License: CC BY-SA 3.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 6, 2016 at 1:10 | comment | added | Saddle Point | Why do we need a scale $\sigma_j$ rather than a covariance matrix which would make the basis function looks like the exponential part of a multivariate Gaussian ? | |
| Oct 17, 2014 at 2:59 | history | bounty ended | Akash Deshpande | ||
| Oct 15, 2014 at 16:06 | comment | added | martino | @O_Devinyak- Many basis expansion methods will require some sort of parameter estimation. There are many ways to find $\mu$ so I don't think this necessarily means we are reducing the problem to SVR. To be honest, I am not an expert on SVR but the loss function that is minimised is certainly different and I am sure many of the features are ignored - thats the Support Vector way. With basis functions we use all the functions for evaluation but luckily compact support means many of the basis functions return neglible or zero values. Anyway,it would make for a good question on this forum | |
| Oct 15, 2014 at 15:25 | comment | added | O_Devinyak | Nice answer! However, searching for $\mu$, don't we finish with support vector machine regression (with gaussian kernel)? | |
| Oct 15, 2014 at 14:13 | history | edited | martino | CC BY-SA 3.0 |
deleted 3 characters in body
|
| Oct 15, 2014 at 13:23 | history | edited | martino | CC BY-SA 3.0 |
deleted 99 characters in body
|
| Oct 15, 2014 at 13:00 | history | edited | martino | CC BY-SA 3.0 |
deleted 12 characters in body
|
| Oct 15, 2014 at 12:51 | history | edited | martino | CC BY-SA 3.0 |
deleted 20 characters in body
|
| Oct 15, 2014 at 12:30 | history | answered | martino | CC BY-SA 3.0 |