Call for Papers: Special Session on Learning from Distributed Data and Knowledge Repositories

Doina Caragea dcaragea at iastate.edu
Tue Feb 26 11:20:00 EST 2002


Call For Papers

SPECIAL SESSION ON
LEARNING FROM DISTRIBUTED DATA AND KNOWLEDGE REPOSITORIES

The 2002 International Conference on Machine Learning and Applications
(ICMLA'02 )
Monte Carlo Resort, Las Vegas, Nevada, USA
June 24-27, 2002


INTRODUCTION

Many practical knowledge discovery tasks present several new challenges in
Machine Learning. The data and knowledge repositories required in these
applications tend to be large, physically distributed, autonomously managed,
and rapidly evolving. Public datasets on the Internet, corporate databases
maintained as a distributed collection of datamarts on the company intranet,
medical data including patient history, repositories containing results of
medical studies and treatment information for the different ailments are
examples of some of the distributed data and knowledge repositories that are
in use today.

Despite the tremendous advances in computing power and communications
infrastructure, the currently well known framework of knowledge discovery
from a centrally located data warehouse is not suitable in several
applications. Accumulating data into a central data warehouse is severely
limited by the available communication bandwidth. Even if the data is
successfully assembled in a central data warehouse, the cost of the
computing infrastructure required to mine such a large volumes of data can
be prohibitive. The rapidly evolving nature of some or all of the data
repositories that feed into data warehouse makes it difficult to keep the
data warehouse up to date. If the distributed repositories are autonomously
maintained then the questions of privacy and security of the data as it is
transferred to a centralized warehouse become crucial.

The scenarios outlined above call for a new distributed learning framework
that should take into account both theoretical aspects and practical
challenges of learning in such environments. There has been a flurry of
activity in the area of learning from distributed data and knowledge
repositories. This technical session is geared to bring together researchers
and practitioners areas such as machine learning, knowledge discovery and
data mining, information extraction, information fusion, software agent
systems and those working on related problems in databases and distributed
computing. It is our hope that this session will facilitate an exchange of
knowledge and ideas and foster further
progress in this interesting and challenging field.


TOPICS

Topics of interest at this special technical session include but are not
limited to:
  a.. Theoretical Foundations:
  Task and data decomposition, knowledge representation, learning operators,
complexity.
  b.. Algorithms:
  Scalable, efficient, robust, parallel and distributed learning algorithms
that can learn from partial schemas, combine multiple models learned on
horizontal or vertical partitions of the data, and update the learned model
quickly and effectively in the presence of every changing data.
  c.. Learning Agents:
  Capable of functioning autonomously, collaborating, communicating, and
co-ordinating the learning task(s) among themselves.
  d.. Architecture:
  Requirements and protocols for data and knowledge representation,
communication  between agents, network bandwidth.
  e.. Privacy and Security:
  The ability to access sensitive information located remotely, the
distribution of  learning tasks among mobile autonomous agents, and the
transmission of learned knowledge between different repositories raises
legitimate security and privacy concerns which must be addressed.
  f.. Applications:
  Challenging new applications in science, engineering, medicine, and
business.

IMPORTANT DATES

Submission Deadline: MARCH 8, 2002
Notification of Acceptance: MARCH 21, 2002
Camera Ready Papers Due: APRIL 22, 2002


SUBMISSION PROCEDURE

An electric version of previously unpublished work at most 6 pages in length
including figures, tables, and references. Electronic versions of the paper
in postscript or PDF format should be submitted via email to Doina Caragea
at dcaragea at iastate.edu.  The first page should mention the Title,
Author(s), Affiliation(s), Contact Author's Name, Mailing address, and
E-mail address.


SESSION CHAIRS

Doina Caragea, Iowa State University (dcaragea at iastate.edu)
Vasant Honavar, Iowa State University (honavar at cs.iatate.edu)
Rajesh Parekh, Blue Martini Software (rparekh at bluemartini.com)
Jihoon Yang, SRA International (yangji at verdi.iisd.sra.com)

All questions and inquiries about the technical session should be directed
to Doina Caragea.



---




More information about the Bioforum mailing list