W. B. Langdon and R. Poli
the contents of your paper is quite intriguing to me; in fact I have noticed
among the keywords words like structural complexity, Occam's razor and MDL
(Minimum Description Length). Moreover, by reading your paper, there is
often implicit reference to the concepts of solution length. These concepts
are quite familiar to me. In fact I am working on the evaluation of
Kolmogorov Complexity (KC) for binary strings
belonging to given grammars; I do not know if you have ever met this idea.
It basically consists in this: given a binary string, the aim is to find
the shortest program (in a given language, say Lisp or its derivative Toy-Lisp)
which terminates its execution and which gives as output that string.
I and my colleagues are solving this by means of Genetic Programming which
helps us to generate new programs from already found ones. The approach
has shown to be working, and the results we are achieving are the first
in this field. (by the way, KC is often also referred to as Solomonoff C. or
Chaitin C.)
I wish to ask you as follows:
- is there in your opinion any link between your work and Kolmogorov
Complexity? It seems, at first glance, that your aim is to find the shortest
"program" able to guide an ant through the trail.
Hoping this can help you and can favour discussion/exchange of ideas
dr. Ivan De Falco