Invited Research Talk 1
Small Device Data Management
Rajkumar Sen
(IIT
ABSTRACT:
Lightweight
computing devices are becoming ubiquitous and increasing number of applications
are being developed for these devices. Many applications e.g
Egovernance, healthcare
etc. need to deal
with a significant amount of data. The queries are quite complicated involving selctions, joins and aggregates. Data processing on a
handheld device involves many complex issues. Applications like Microbanking involve transactional details. Hence,
atomicity and durablity properties need to be
satisfied. All such issues necessitate the presence of a local database
management system on the device. Also, a DBMS will make application code
development easy and fast. However, scaling down the database has not been easy
as these devices are constrained by limited stable storage and main memory.
Resources in handhelds are increasing with every new device; optimum
utilization of these limited resources is a must for such a database system.
Evaluation of complex joins and aggregates requires efficient usage of the
limited main memory. Memory should be optimally allocated among the database
operators. We present in this talk a new data storage model, ID Based Storage
which reduce storage cost considerably. We present an exact and a heuristic
method for optimally allocating memory among the database operators and compare
both the methods. Our query processing strategy guarantees the best query
execution plan for any handheld device depending on the available resources.
Our lightweight DBMS for handheld devices has been tested on the Simputer.
SPEAKER
BIO:
Rajkumar Sen received his MTech degree from Dept. of Computer Science &
Engineering, IIT Bombay, in 2003. The work presented here is part of his
master’s thesis and will appear in the forthcoming IEEE ICDE 2005 conference.
Invited
Research Talk 2
Privacy-Preserving Data Mining
Shipra Agrawal
(IISc Bangalore and
ABSTRACT:
To preserve client privacy in the data mining
process, a variety of techniques based on random perturbation of individual
data records have been proposed recently.
In this talk, we present FRAPP, a generalized matrix-theoretic framework
of random perturbation, which facilitates a systematic approach to the design
of perturbation mechanisms for privacy-preserving mining. Specifically, FRAPP is used to demonstrate
that (a) the prior techniques differ only in their choices for the perturbation
matrix elements, and (b) a symmetric perturbation matrix with minimal condition
number can be identified, maximizing the accuracy even under strict privacy
guarantees. We also propose a novel perturbation mechanism wherein the matrix
elements are themselves characterized as random variables, and demonstrate that
this feature provides significant improvements in privacy at only a marginal
cost in accuracy.
SPEAKER
BIO:
Shipra Agrawal graduated with an ME
degree from Dept. of Computer Science & Automation, Indian Institute of
Science,