SGD is sequential, but if you have millions of examples to classify, you can run many copies of the SGD model in parallel. This is a trivial form of parallelism, but it is often encountered in practice.
On Tue, Jun 25, 2013 at 5:31 PM, Suneel Marthi <[email protected]>wrote: > ... and SGD is sequential. > > > > > ________________________________ > From: "Chandra Mohan, Ananda Vel Murugan" <[email protected]> > To: "[email protected]" <[email protected]> > Sent: Tuesday, June 25, 2013 12:05 PM > Subject: RE: Parallel implemented classifiers in Mahout > > > As far as I know, Random Forest, Naïve Bayes and complimentary Naïve Bayes > are parallel algorithms for classification. > > -----Original Message----- > From: Bill Q [mailto:[email protected]] > Sent: Tuesday, June 25, 2013 9:30 PM > To: [email protected] > Subject: Parallel implemented classifiers in Mahout > > Hi, > I am trying to get a better understanding of the classification algorithms > that the training part has been implemented parallel so it can > run simultaneously on different computing nodes. > > I have tried to read some of the documentation but I got a little confused > of the wording. For example, the Logistic SGD, is it implemented parallel? > And SVM? Can some veteran here give a list of the the algorithm that can be > trained in a distributed parallel way? > > Many thanks. > > > Bill >
