Genetic algorithms[ John R. Koza, Martin A. Keane and Matthew J. Streeter, ``Evolving Inventions'', < I> Scientific American< /I> , vol. 288, no. 2, Feb. 2003< /A> ]

John R. Koza, Martin A. Keane and Matthew J. Streeter, ``Evolving Inventions'', Scientific American, vol. 288, no. 2, Feb. 2003] mimic biological processes of evolution ; they can be used to find solutions in domains where:

• A solution consists of values for several independent variables.

• Some variables may have discrete values; others may have continuous numerical values.

• A candidate solution can be evaluated by some evaluation function.

Example: An animal might have variables color, percent-body-fat, length-of-legs.

It would be possible to use search on the discrete variables and hill-climbing on the continuous variables, but often the search space is too large.

Contents    Page-10    Prev    Next    Page+10    Index