[Rd] knnFinder package

Christos Hatzis christos.hatzis at nuverabio.com
Fri Jan 18 18:15:08 CET 2008


Hello,

Perhaps not directly related to knnFinder package, but since the discussion
is on knn-related algorithms, has anyone looked into incorporating the cover
tree algorithm that was invented by the Yahoo team recently?  It is supposed
to be remarkably fast for exact kNN computations in very large datasets.

The link below has details on the algorithm and a C++ implementation from
the inventor.  I don't know enough C++ to be able to work through the
wrapper myself, so I wanted to pass this on to the list in case there is
interest.

http://hunch.net/~jl/projects/cover_tree/cover_tree.html

All the best,

-Christos

Christos Hatzis, Ph.D.
Nuvera Biosciences, Inc.
400 West Cummings Park
Suite 5350
Woburn, MA 01801
Tel: 781-938-3830
www.nuverabio.com
 

> -----Original Message-----
> From: r-devel-bounces at r-project.org 
> [mailto:r-devel-bounces at r-project.org] On Behalf Of Andrew Finley
> Sent: Friday, January 18, 2008 10:20 AM
> To: Roger.Bivand at nhh.no; gpau at ebi.ac.uk
> Cc: r-devel at r-project.org
> Subject: Re: [Rd] knnFinder package
> 
> Hello all,
> The ann() function in the yaImpute package provides a more 
> full featured interface to the same underlying approximate 
> nearest neighbor library
> (http://www.cs.umd.edu/~mount/ANN) that is called by the nn() 
> function in knnFinder. knnFinder is actually calling an older 
> (defunct) version of the ANN library which was known to cause 
> some segmentation faults.
> 
> I would be willing to adopt knnFinder, but it is now 
> redundant with the functions offered in yaImpute. But first a 
> question -- Should we maintain existing packages that have 
> been superseded by other packages? 
> If so, then I'll adopt knnFinder, but if not then I suggest 
> letting it go.
> 
> Kind regards-
> Andy
> 
> 
> On 14 Jan 2008, Roger Bivand wrote:
>  > On Mon, 14 Jan 2008, Gregoire Pau wrote:
>  >
>  > > Dear all,
>  > >
>  > > I have found some serious bugs in the knnFinder package 
> (which supports  > > data structures and algorithms for both 
> exact and approximate nearest  > > neighbor searching in 
> arbitrarily high dimensions) that may trigger  > > 
> segmentation faults.
>  > > I have fixed them but I had troubles to contact its 
> maintainer Samuel E.
>  > > Kemp (previously with the University of Glamorgan, UK). 
> Do someone know  > > where I can reach him ?
>  >
>  > I have also tried, for the same reason, without result. He 
> seems to have  > been a student who has now left the 
> university where he wrote the package.
>  > I also wrote to his apparent supervisor, also without 
> result. My  > suggestion would be that the package be 
> orphaned, and that if you are  > willing to replace it with 
> your version, and to adopt it as maintainer,  > its posting 
> on CRAN can be continued.
>  >
>  > Roger
>  >
>  > >
>  > > Best regards,
>  > >
>  > > Greg
>  > > ---
>  > > Gregoire Pau
>  > > EMBL/EBI Cambridge, UK
>  > > http://www.ebi.ac.uk/~gpau
>  > >
>  > > ______________________________________________
>  > > R-devel at r-project.org mailing list
>  > > https://stat.ethz.ch/mailman/listinfo/r-devel
>  > >
>  >
>  > --
>  > Roger Bivand
>  > Economic Geography Section, Department of Economics, 
> Norwegian School of  > Economics and Business Administration, 
> Helleveien 30, N-5045 Bergen,  > Norway. voice: +47 55 95 93 
> 55; fax +47 55 95 95 43  > e-mail: Roger.Bivand at nhh.no  >  > 
> ______________________________________________
>  > R-devel at r-project.org mailing list
>  > https://stat.ethz.ch/mailman/listinfo/r-devel
> 
> 
> --
> Andrew Finley, PhD
> Natural Resources Building
> Michigan State University
> East Lansing, MI 48824-1222
> Phone: 517-432-7219
> Fax: 517-432-1143
> web: http://blue.for.msu.edu
> 
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
> 
>



More information about the R-devel mailing list