2012/9/5 Nelle Varoquaux <[email protected]>
>
>
> On 5 September 2012 08:08, Matthieu Brucher <[email protected]>wrote:
>
>> Excellent work!
>> I have a question on MDS. Is it the classic MDS or something else?
>> (Asking the question as PCA is the classic MDS). It seems to be when the
>> distance matrix is Euclidean?
>>
>
> It is indeed the classical MDS. When the whole similarity matrix, indeed,
> the classical MDS is equivalent to a PCA. But when there a missing values,
> PCA don't work.
> An improvement to do to the current implementation is to use the PCA when
> the whole matrix is provided. That should speed it up a lot.
>
> Cheers,
> N
>
Hi Nelle,
Thanks for the intel, I forgot it could handle sparse matrices.
One additional question: is it possible to have details on the minimized
cost function? As there are thousands of different possibilities, sometimes
dealing with outiliers (ans so shortcuts), small values...
Cheers,
Matthieu
--
Information System Engineer, Ph.D.
Blog: http://matt.eifelle.com
LinkedIn: http://www.linkedin.com/in/matthieubrucher
Music band: http://liliejay.com/
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
