abstract = "We present an approach for developing parallel
distributed implementation of genetic programming
(PDIGP) based on exploitation of the inherent
parallelism among semi-isolated subpopulations.
Proposed implementation runs on cost-efficient
configurations of clusters in LAN and/or Internet
environment. PDIGP features single global migration
broker and centralized manager of the semi-isolated
subpopulations, which contribute to achieving quick
propagation of the globally fittest individuals among
the subpopulations, reducing the performance demands to
the communication network, and achieving flexibility of
system configurations by introducing dynamically
scaling up opportunities. PDIGP exploits distributed
component object model (DCOM) as a communication
paradigm, which offers generic support for the issues
of naming, locating and protecting the distributed
entities in PDIGP. Experimentally obtained results show
that in some system configurations the computational
effort is less than the computational effort in
canonical panmictic GP. Analytically obtained and
empirically proved results of the speedup of the
computational performance indicate that PDIGP features
linear, close to ideal characteristics, which, together
with the observed reduction of the computational effort
contribute to the acquaintance of hyper-linear overall
speedup in developed PDIGP.",
notes = "symbolic regression, many node distributed
computing
Muroran Institute of Technology Mizumoto 27-1,
Muroran,JAPAN 050-8585",
