mahout-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Samee Zahur (JIRA)" <j...@apache.org>
Subject [jira] Commented: (MAHOUT-24) Skeletal LWLR implementation
Date Fri, 11 Apr 2008 06:22:04 GMT

    [ https://issues.apache.org/jira/browse/MAHOUT-24?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12587860#action_12587860
] 

Samee Zahur commented on MAHOUT-24:
-----------------------------------

Actually I do have some Unit tests in there, albeit rudimentary. And the reason I didn't comment
it was, like I said, it is a skeletal implementation, which might need to change significantly
before being commited.

As for mapping the concepts of the nips paper, it simply calculates the sum(x[i]x[j]) mentioned
in the nips paper, nothing else. So to the actual set of regression equations is this:

x[i] = x[ind] * output[i]/output[ind]

where ind is the index of the independent variable, x[ind] is the test input at which we want
to use the regression equation for predictions, and x[i] is the ith component of the predicted
output. output[i] represents the ith line in the output file.

For example, in the 2D case, the input file is supposed to contain a list of points along
with their weights. If the output file results in something like this:
0: 2
1: 4

then the regression equation is 

y = x*4/2
or, y=2x

where 4=output[0], 2=output[1], x is x[ind]=x[0], y is x[1].

When I complete the implementation, I will have all the documentations ready to clear this
up. But first

Samee

> Skeletal LWLR implementation
> ----------------------------
>
>                 Key: MAHOUT-24
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-24
>             Project: Mahout
>          Issue Type: New Feature
>         Environment: n/a
>            Reporter: Samee Zahur
>         Attachments: LWLR.patch.tar.bz2
>
>
> This is a very skeletal but functional implementation for LWLR. It outputs n lines where
n is the number of dimensions. ith line = sum(x[i]*x[ind]) where ind is the index of independant
variable. So the actual gradient = 2nd line/1st line for the classical 2D.
> Contains a single small test case for demonstration.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message