On Thu, Jun 9, 2011 at 2:27 AM, Lance Norskog wrote:
> Projecting to the first "two" singular vectors?
Yes.
> Do an SVD on a random matrix, and use the first 2 (or three) singular
> vectors as a matrix?
Not a random matrix. A matrix of positions shifted back to have
average mean (aka PCA).
>
Just tested seq2sparse using binary distribution again and received:
11/06/08 21:17:00 INFO mapred.JobClient: Task Id :
attempt_201106061352_0066_r_01_1, Status : FAILED
Error: java.lang.ClassNotFoundException: org.apache.lucene.analysis.TokenStream
at java.net.URLClassLoader$1.run(URLCl
Projecting to the first "two" singular vectors?
Do an SVD on a random matrix, and use the first 2 (or three) singular
vectors as a matrix?
What goes into the affinity matrix?
On Wed, Jun 8, 2011 at 4:24 PM, Ted Dunning wrote:
> Projecting to the first to singular vectors is better.
>
> Forming a
I explained in an earlier post that I was having problems running some
examples on a cluster when using the binary distribution. My cluster was
complaining about missing classes.. ie lucene analyzer and google
preconditions. However when I tried the same thing on a src distribution
(and after m
Hello all,
I am trying to run seq2sparse as follow:
bin/mahout seq2sparse \
-i clustering/items-seq \
-o clustering/items-vectors \
-wt tfidf \
-nr 3 \
-ng 3 \
-s 5 \
-md 3 \
-x 90 \
-ml 50 \
-ow
The first tas
Projecting to the first to singular vectors is better.
Forming an affinity (rather than distance) matrix and projecting to
those coordinations is very interesting.
On Thu, Jun 9, 2011 at 12:25 AM, Lance Norskog wrote:
> I've used multi-dimensional scaling (MDS) in another toolkit to
> down-proje
I got a slightly different error on the next line of KMeansDriver.java
(running on OS X Snow Leopard)
11/06/08 16:02:12 INFO compress.CodecPool: Got brand-new compressor
Exception in thread "main" java.lang.ClassCastException:
org.apache.hadoop.io.IntWritable cannot be cast to
org.apache.mahout.ma
I've used multi-dimensional scaling (MDS) in another toolkit to
down-project high-dim vectors to 2d and 3d. What tools for this are
available in Mahout? Random Projection down to 2 dimensions is easy,
but seems unsound.
--
Lance Norskog
goks...@gmail.com
I would encourage you to take a stab at a patch on this. You aren't
the only person to have expressed interest in scaling PCA, but you
aren't a member of a large horde, either.
On Wed, Jun 8, 2011 at 7:39 AM, Eshwaran Vijaya Kumar
wrote:
> Thanks Ted. That is good news.
> On Jun 7, 2011, at 11:1
i guess the only problem that creates such demand for CDH is the fact
that hadoop project twisted everybody's arm by deprecating the entire
MR api over what seems to be just perceived OOA design issues but not
functional issues. Even that would've been ok if it weren't for the
fact that they did no
Have a look at the Classify class in the classifier package as a starting place.
-Grant
On Jun 8, 2011, at 4:32 AM, 刘逸哲 wrote:
> Hi all，
> There are trainclassifier and testclassifier, but I would like to
> know how to make prediction on new text with any lable.
> I think the te
Thanks Ted. That is good news.
On Jun 7, 2011, at 11:12 PM, Ted Dunning wrote:
> I think that incorporating mean subtraction into the SSVD code should
> be relatively straightforward. The trick is that you have to project
> the orginal matrix and the mean separately and then combine the
> results
Hi all，
There are trainclassifier and testclassifier, but I would like to know
how to make prediction on new text with any lable.
I think the testclassifier must make a prediction at first, but the
testclassifier interface need input texts with lables?
Is there an easy
Hadoop is Hadoop, so I don't know that any roadmap inconsistency between CDH
and Hadoop is somehow Hadoop's fault.
I don't think it's this ambiguous. Mahout runs on 0.20.2 Amazon EMR runs
0.20.2. The latest Hadoop version is 0.20.203.0. CDH is indeed somewhere
inbetween but that's CDH.
On Wed, Ju
14 matches
Mail list logo
