UTCS Colloquia/AI - Jeff Schneider/Carnegie Mellon University - Robotics Institute, "Learning Dynamic Models with Non-sequenced Data", ACES 2.302

Contact Name: 
Jenna Whitney
Date: 
Oct 28, 2011 11:00am - 12:00pm

There is a sign-up schedule for this event that can be found at

http://apps.cs.utexas.edu/talkschedules/cgi/list_events.cgi

Type o

f Talk: UTCS Colloquia/AI

Speaker/Affiliation: Jeff Schneider/Carnegie
Mellon University - Robotics Institute

Talk Audience: UTCS Faculty,

Graduate Students, Undergraduate Students, and Outside Interested Parties

Date/Time: Friday, October 28, 2011, 11:00 a.m.

Location: ACE

S 2.302

Host: Joydeep Ghosh/Co-sponsored by Yahoo Data Mining Series

n
Talk Title: Learning Dynamic Models with Non-sequenced Data

Tal

k Abstract:
Virtually all methods of learning dynamic systems from data s

tart with
the same basic assumption: the learning algorithm will be given
a time
sequence of data generated from the dynamic system. We consider t

he case
where the training data comes from the system’s operation bu

t with no
temporal ordering. The data are simply drawn as individual disc

onnected
points. While making this assumption may seem absurd at first gl

ance, we
observe that many scientific modeling tasks have exactly this p

roperty.

We propose several methods for solving this problem. We write
down an
approximate likelihood function that may be optimized to learn d

ynamic
models and show how kernel methods can be used to obtain non-linea

r
models. We propose an alternative method that focuses on achieving
te

mporal smoothness in the learned dynamics. Finally, we consider the
case
where a small amount of sequenced data is available along with a
large a

mount of non-sequenced data. We propose the use of the Lyapunov
equation

and the non-sequenced data to provide regularization when
performing regr

ession on the sequenced data to learn a dynamic model. We
demonstrate our
methods on synthetic data and describe the results of our
analysis of so

me bioinformatics data sets.

Speaker Bio:
Dr. Jeff Schneider is an a

ssociate research professor in the Carnegie
Mellon University School of C

omputer Science. He received his PhD in
Computer Science from the Univers

ity of Rochester in 1995. He has over 15
years experience developing, pu

blishing, and applying machine learning
algorithms in government, scien

ce, and industry. He has dozens of
publications and has given numerous i

nvited talks and tutorials on the
subject.

In 1995 Dr. Schneider co-

founded and became CEO of Schenley Park
Research, a company dedicated to
bringing new machine learning algorithms
to industry. In 2004 he develop

ed a new machine-learning based CNS drug
discovery system and spent two y

ears as the CIO of Psychogenics to
commercialize the system. Through his

academic, commercial, and
consulting efforts, he has worked with sever

al dozen companies and
government agencies around the world.