Mining, Decision Support, and Knowledge Management: Applying
hBOATM to data mining, information
visualization, decision support, and knowledge management will lead
to better decision making and resource utilization.
Design and Optimization: This technology can optimize the
design of engineering systems in a variety of disciplines. Aerospace,
mechanical, civil, environmental, chemical, bioengineering, telecommunications,
electronic, and computer products, processes, and systems can all employ
hBOATM to improve system performance,
cost and efficiency.
Supply-Chain, Operations Research, and Enterprise Resource Management:
Optimizing time and resource utilization in large-scale businesses
and organizations increasingly relies on sophisticated solution algorithms
coupled with sophisticated enterprise-wide software systems. hBOATM
can augment or supplant first-generation solution techniques to provide
higher quality time plans and resource allocation.
combines techniques from genetic algorithms and evolutionary computation
with the latest in Bayesian networks and artificial intelligence to
create a powerful solver that discovers and exploits unknown irregularities
in search, optimization, and machine learning problems. Specifically,
points are sampled initially at random,
highly fit subset of points is modeled using Bayesian networks,
network searches, and appropriate theoretical or Bayesian metrics
Bayesian network model is sampled to generate a population of new
process is iterated by population, and the resulting procedure quickly
uncovers unknown or hidden regularities and directs the search to solutions
of highest quality in times that often grow no more than a quadratic
function of the number of decision variables. Heirarchy is handled
through the addition of niching techniques and compact data structures
that facilitate the inheritance of building blocks as whole chunks.
The bottom line is a broadly competent procedure that solves enormously
difficult problems (problems with millions or billions of optima) with
advantages for hBOATM
can solve previously intractable problems today, but this technology
is not yet widely available in commercial software packages or systems.
Software purveyors or end users who adopt hBOATM
now will be able to acheive higher quality solutions to larger problems
more quickly than their competitors.
should be of particular interest to:
suppliers, particularly those who embed search, optimization, or machine
learning in their product offerings will want to augment or replace
existing tools with hBOATM. Sophisticated
end users who perform or would like to perform optimization using existing
in-house analysis codes can easily add hBOATM
and quickly provide broadly competent optimization capability.