Indeed, in Coates' code the bias term is not penalized.
Is there any way to turn off the bias penalty in liblinear?

On Wed, Feb 15, 2012 at 3:57 AM, Paolo Losi <[email protected]> wrote:
>
>
> On Wed, Feb 15, 2012 at 8:26 AM, Olivier Grisel <[email protected]>
> wrote:
>>
>> 2012/2/15 Ian Goodfellow <[email protected]>:
>> > Further update: I talked to Adam Coates and his code doesn't implement
>> > a standard SVM. Instead it's an "L2 SVM" which squares all the slack
>> > variables. So this probably explains the difference in performance I
>> > observed prior to building this test case.
>>
>> Good to know. AFAIK this is the same loss (squared hinge) as used by
>> default in liblinear. You could try to compare the outcome with
>> LinearSVC (albeit you will get the memory copy of you dense input) or
>> SGDClassifier (the penalty and loss parameters will allow you to
>> adjust the objective function to your will).
>
>
> ... and still those (liblinear vs Adam Coates) are not exactly comparable
> since
> the bias term is penalized in liblinear while in Coates's I suppose it's
> not.
>
> Paolo
>
>
> ------------------------------------------------------------------------------
> Virtualization & Cloud Management Using Capacity Planning
> Cloud computing makes use of virtualization - but cloud computing
> also focuses on allowing computing to be delivered as a service.
> http://www.accelacomm.com/jaw/sfnl/114/51521223/
> _______________________________________________
> Scikit-learn-general mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
>

------------------------------------------------------------------------------
Virtualization & Cloud Management Using Capacity Planning
Cloud computing makes use of virtualization - but cloud computing 
also focuses on allowing computing to be delivered as a service.
http://www.accelacomm.com/jaw/sfnl/114/51521223/
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to