ABSTRACT
Complex architectures can be decomposed into their constituents
that can be straightforward, as well as analytically tractable, and
reconstructed in reverse, yielding a holistic representation that
is fully descriptive in an elegant and uniform manner using
mathematical modeling tools. This kind of approach is what Prof.
Anthony G. Constantinides inspired the author to explore when
he was under his supervision during the PhD period at Imperial
(1994-1997). Since then, the author has engaged in a number of
topics-graph theoretic representation of image data, data pruning,
models of learning by neural networks, and proposal of artificial
mind system. The Festschrift reviews the author’s research activities
so far relevant to modeling complex systems, with a reminiscence of
the days in the past, and suggests some future directions.