The Global War on Terror (GWOT) and the need for Maritime
Domain Awareness (MDA) has led to the need to synthesize a
variety of necessarily disparate data types. These data
types can include representations such as text (email
transmissions, ship logs, computer logs), audio
transmissions (speech, cell phone, and other intercepted
communications), imagery (UAV imagery, satellite imagery;
multispectral imagery, hyperspectral imagery) fingerprint
data, facial recognition data, and other abstract data
types such as ship structural plans or network data. One
would like to be able to use this data for a variety of
purposes, including inference, prediction, synthesis, and
general data mining. The hope is that through the fusion of
available disparate data types one may obtain superior
performance in the aforementioned exploitation tasks.
A prerequisite to the execution of any of these tasks is
the measurement of the similarity or dissimilarity of the
various data types; this is no simple task given the fact
that the data types reside in spaces of differing
dimensionality and structure.
The goal of the proposal is to develop new
mathematical/statistical/computational strategies for the
measurement of similarity or dissimilarity on these
disparate data types through the embedding and exploitation
of such disparate data.