My research is concerned with developing nonlinear models that describe
the complex relationship between the solar magnetic field and the Earths
magnetic field. Building computational models of this relationship can
lead to the development of advanced warning systems for geomagnetic
storms. This is of critical importance to technological systems that
society heavily depends on.
Currently I am focusing on Computational Intelligence based models of
geomagnetic storms. One of the main computational intelligence based model
in use in geomagnetic storm modeling is the recurrent neural network
(RNN). I, along with my research collaborator (Dr Derrick Mirikitani) have
developed sophisticated training algorithms for the RNN which allow for
optimization of model priors and parameters for accurate prediction of Dst
index , and the ability to forecast Dst and update model parameters
online.
Future research will look into the performance of evolutionary algorithms
for better parameterization of models as well as evolving neural
architectures for Dst prediction.
My Previous research interests were within the areas of
program slicing, the semantics of program slicing and program transformation.
I am part of the Program
Transformation and Analysis Group
.
|
Useful links