[jira] [Commented] (MADLIB-938) Non-linear SVM - polynomial kernel

["Xiaocheng Tang (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/MADLIB-938?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15088153#comment-15088153 ] 

Xiaocheng Tang commented on MADLIB-938:
---------------------------------------

The approach is described in [Kar and Karnick|http://jmlr.csail.mit.edu/proceedings/papers/v22/kar12/kar12.pdf] (see Algorithm 1 Random Maclaurin Feature Maps), which made use of the Maclaurin series expansion to approximate inner product kernels (with polynomial kernel as a special case). 

> Non-linear SVM - polynomial kernel
> ----------------------------------
>
>                 Key: MADLIB-938
>                 URL: https://issues.apache.org/jira/browse/MADLIB-938
>             Project: Apache MADlib
>          Issue Type: New Feature
>          Components: Module: Support Vector Machines
>            Reporter: Frank McQuillan
>            Assignee: Xiaocheng Tang
>             Fix For: v1.9
>
>
> Story
> As a MADlib developer, I want to implement a polynomial kernel in the SVM algorithm, so that I get better classification/regression results for certain types of data sets.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)