RESEARCH INTERESTS Research Home Learning Systems Syntactical Pattern Recognition Statistical Pattern Recognition Adaptive Data Structures Artificial Neural Networks Database Systems Data Compression Robotics Data Retrieval and Storage

Following are some of the major results obtained in the area Database Systems. Have enhanced micro-aggregation for secure statistical databases by (a) Using dependence-based information (PAA 2013, Proc. AI 2008), (b) Using the concepts of "Association" and "Interaction" (IEEE T:SMC 2010, Proc. ICICS 2007), (c) Utilizing fixed-structure learning automata (IEEE T:SMC 2009, Proc. PSD 2006), and (d) Optimizing k-Ward techniques (Proc. ACISP 2006). The ACMs, a new patented histogram-based technology, (see below) have been successfully incorporated as a prototype into ORACLE. (LNBIP-3 2008, Proc. ICEIS 2006). This talk was a Plenary/Keynote Talk at the Conference. Have proposed methods to enhance caching in distributed databases by using intelligent Polytree representations (Proc. CanAI 2003). Have developed new histogram-like methods, known as Attribute Cardinality Maps, to approximate the underlying data distribution of an attribute value. ACMs been shown to be much more accurate for query result size estimation than the traditional histogram methods (IEEE T:SMC 2003, Proc. IDEAS 1999, 2000). The patent for this was granted in March 2005. The fact that the ACM histograms would give better query evaluation plans was demonstrated formally (The Comp. Journal 2002). Have developed two novel strategies to obtain (near) optimal ACMs that significantly improve the accuracy of query result size estimation. Have formally derived the relationship between the accuracy of a histogram method and the efficiency of the query evaluation plan that it yields (Proc. DASFAA 2001).