[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Principal components analysis writeup
- To: Dan.Bryce@asu.edu
- Subject: Re: Principal components analysis writeup
- From: Subbarao Kambhampati <rao@asu.edu>
- Date: Tue, 06 Feb 2001 11:17:59 -0700
- Cc: rao@asu.edu, cse494-s01@asu.edu
- In-reply-to: "Your message of Tue, 06 Feb 2001 11:13:01 -0700 (MST)"<Pine.GSO.4.21.0102061111280.984-100000@general1.asu.edu>
- References: <Pine.GSO.4.21.0102061111280.984-100000@general1.asu.edu>
You should have 5 distinct page images,
although pdf file has 8 pages. Ignore the empty pages.
Rao
From: Dan Bryce <Dan.Bryce@asu.edu>
Subject: Re: Principal components analysis writeup
Date: Tue, 06 Feb 2001 11:13:01 -0700 (MST)
Message-ID: <Pine.GSO.4.21.0102061111280.984-100000@general1.asu.edu>
Dan.Bryce> Hi,
Dan.Bryce>
Dan.Bryce> I just printed the document and a few of the pages have nothing on them. Are
Dan.Bryce> the pages missing data, or are they just extraneous?
Dan.Bryce>
Dan.Bryce> Thanks,
Dan.Bryce>
Dan.Bryce> Dan Bryce
Dan.Bryce>
Dan.Bryce>
Dan.Bryce>
Dan.Bryce> On Mon, 5 Feb 2001, Subbarao Kambhampati wrote:
Dan.Bryce>
Dan.Bryce> >
Dan.Bryce> > The link
Dan.Bryce> >
Dan.Bryce> > http://rakaposhi.eas.asu.edu/cse494/notes/pplcomps.pdf
Dan.Bryce> >
Dan.Bryce> > contains a very nice writeup on principal component analysis--which is
Dan.Bryce> > the theory behind LSI style dimensionality reduction techniques.
Dan.Bryce> >
Dan.Bryce> > The write-up is taken from Christopher Bishop's "Neural Networks for
Dan.Bryce> > Pattern Recognition". The picture of the ellipise shaped data I showed
Dan.Bryce> > in the class was from here..
Dan.Bryce> >
Dan.Bryce> > This write up explains very elegantly exactly why eigen-vectors wind
Dan.Bryce> > up getting mixed up in dimensionality reduction. It also points out
Dan.Bryce> > that dimensionality reduction using eigen vectors is a "linear"
Dan.Bryce> > method--in that it considers linear translation and rotation of
Dan.Bryce> > coordinate frames. It doesn't support nonlinear transformations (see
Dan.Bryce> > the little cicle picture).
Dan.Bryce> >
Dan.Bryce> > Here is hoping that someone reads this stuff and redeems the hour I
Dan.Bryce> > spent scanning the pages, converting them to jpegs, including them in
Dan.Bryce> > a word file, converting the word file to postscript and then to pdf,
Dan.Bryce> > and uploading the pdf.
Dan.Bryce> >
Dan.Bryce> > Rao
Dan.Bryce> > [Feb 5, 2001]
Dan.Bryce> >
Dan.Bryce> >
Dan.Bryce> > ps: What did you say--give you xerox copies? Aww..that is so
Dan.Bryce> > outmoded--look at all the time I saved by not having to go to the
Dan.Bryce> > xerox machine.
Dan.Bryce> >
Dan.Bryce> >
Dan.Bryce>