SPECIAL SESSION: "Multidimensional Visualization and Some Applications”
(Subtitle: “How to become a Multidimensional Detective”)
Clinical Center (Building 10)
CRC Room 2-3330
- view the seminar archive
DESCRIPTION: (by Alfred
Inselberg) A dataset with M items has 2^M subsets anyone
of which may be the one we really want. With a good data
display, our fantastic pattern-recognition ability can
not only cut great swaths searching through this
combinatorial explosion but also extract insights from
the visual patterns. These are the core reasons for data
visualization. With parallel coordinates (abbr. ||-coords)
the search for multivariate relations in high
dimensional datasets is transformed into a 2-D pattern
recognition problem. Guidelines and strategy for
knowledge discovery are illustrated on different real
datasets, some with hundreds of variables. A geometric
classification algorithm based on ||-coords is presented
and applied to complex datasets. It has low
computational complexity providing the classification
rule explicitly and visually. The minimal set of
variables required to state the rule is found and
ordered by their predictive value. A visual economic
model of a real country is constructed and analyzed to
illustrate how multivariate relations can be modeled by
means of hypersurfaces - discovering interrelations,
sensitivies, feasible states and doing trade-offs. For
another application, Automatic Collision Avoidance
algorithms for Air Traffic were developed (3 patents).
An overview provides foundational understanding of ||-coords
and a prelude of what is on the way: the concentration
of the relational information into clear patterns
eliminating visual clutter altogether. (Do not be
intimidated by this formal description. The speaker is
also well known for his numerological anecdotes and
palindromic digressions!)
COMMENTARY: (by Jim DeLeo)
I see data visualization as a fundamental step in
biomedical data mining. With it, the investigator can
“play” with data like a child playing a video game, and
in so doing perhaps get insights that suggest
significant hypotheses worthy of exploring further with
other data mining tools. We (BCIG) have explored
parallel coordinates in the past, and I have continued
to be a big activist in encouraging folks to consider
this methodology in basic and clinical biomedical
research projects. At the present time, I am
particularly interested in collaborating with anyone
associated with the NIH who has clinical data in the
CRIS system and who might like to explore using parallel
coordinates to examine their data. If you are such a
person, please contact me (jdeleo@nih.gov).
REGISTRATION: As with all BCIG
events, registration is not required. Just show up
happy.
WEBCASTING: This event will be
web cast live and be made available for post program
viewing on the BCIG web site (www.nih-bcig.org).
To get more information about our webcasting service,
please contact Meeting Master Carl Leonard by e-mail:
cleonard@lired.com
or by calling him on 301-496-0191. NIH CONTACT: Jim
DeLeo, 301-496-3848,
jdeleo@nih.gov
REFRESHMENTS: Bring refreshments
if you would like. There is an open cafeteria near the
meeting room.
BCIG WEB SITE:
www.nih-bcig.org
NIH VISITOR INFORMATION:
http://www.nih.gov/about/visitor/
|
3:00 pm - 4:30 pm October 31, 2006
SPEAKER: Alfred Inselberg received a Ph.D. in Applied Mathematics and
Physics (1965) from the University of Illinois (C-U campus). He was employed by
the IBM Scientific Centers and Research in the USA for many years, and held
academic positions at the University of Illinois, UCLA, USC, Technion and
Ben-Gurion University in Israel. In 1996, he was elected Senior Fellow at the
San Diego Supercomputing Center. Since 1977, he has been developing the
multidimensional visualization methodology which he invented and became known as
"Parallel Coordinates". He has authored numerous technical papers, has several
patents and awards. As of 1995 he teaches at Tel Aviv University and founded his
own company MDG providing services and software in Visualization, Exploration
and Data Mining of high dimensional sets. Currently he is completing the
textbook on Parallel Coordinates (Springer) ... and hi-tech entertainment. |
