[R] Use of nnet() with package SNOW ?

From: Ron Piccinini <ronpicci_at_yahoo.fr>
Date: Sun 04 Jul 2004 - 12:11:25 EST


Hello R masteRs,

I am trying to train a neural network whose training set is about 910,000 rows by 5 columns. I am running R on a small cluster of 8 machines (7 slaves) using the SNOW package. Is there a smart way to use all 8 processors to train the neural net? or am I just better off putting all the RAM possible on one machine and run nnet on one processor? I thought about storing each column of the training set on a seperate node and have the master node execute something of the sort: nnet((cbind(clustr[[1]],clustr[[2]],...,clustr[[5]])),targetvector,,5)

(where clstr is the name of the contents of the cluster (is diferent from "cl")
However I think that when I do that R loads the cbind-ed matrix in the master's node environment and therefore no parallel processing gains are realized. Thanks in advance for your comments and suggestions,

Ron.



R-help@stat.math.ethz.ch mailing list
https://www.stat.math.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html Received on Sun Jul 04 12:16:05 2004

This archive was generated by hypermail 2.1.8 : Wed 03 Nov 2004 - 22:54:40 EST