Q4: How many EAs exist? Which?

 The All Stars
     There  are  currently  3  main  paradigms  in  EA  research:  GENETIC
     ALGORITHMs,   EVOLUTIONARY  PROGRAMMING,  and  EVOLUTION  STRATEGIEs.
     CLASSIFIER SYSTEMs and GENETIC PROGRAMMING are OFFSPRING  of  the  GA
     community.   Besides  this  leading  crop,  there  are numerous other
     different approaches, alongside hybrid experiments, i.e. there  exist
     pieces  of software residing in some researchers computers, that have
     been described in papers in conference proceedings, and  may  someday
     prove  useful  on certain tasks. To stay in EA slang, we should think
     of these evolving strands as BUILDING BLOCKs,  that  when  recombined
     someday,  will  produce  new  offspring  and  give  birth  to  new EA
     paradigm(s).

 Promising Rookies
     As far as "solving complex function  and  COMBINATORIAL  OPTIMIZATION
     tasks"  is  concerned, Davis' work on real-valued representations and
     adaptive operators should be mentioned (Davis 89). Moreover Whitley's
     Genitor  system  incorporating  ranking  and "steady state" mechanism
     (Whitley   89),   Goldberg's   "messy   GAs",    involves    adaptive
     representations (Goldberg 91), and Eshelman's CHC algorithm (Eshelman
     91).

     For  "the  design  of  robust  learning  systems",  i.e.  the   field
     characterized  by  CFS, Holland's (1986) CLASSIFIER SYSTEM, with it's
     state-of-the-art implementation CFS-C  (Riolo  88),  we  should  note
     recent  developments  in  SAMUEL  (Grefenstette 89), GABIL (De Jong &
     Spears 91), and GIL (Janikow 91).

     References

     Davis,  L.  (1989)  "Adapting  operator  probabilities   in   genetic
     algorithms", [ICGA89], 60-69.

     Whitley,  D.  et  al.  (1989)  "The  GENITOR  algorithm and SELECTION
     pressure: why rank-based allocation of reproductive trials is  best",
     [ICGA89], 116-121.

     Goldberg,  D. et al. (1991) "Don't worry, be messy", [ICGA91], 24-30.

     Eshelman, L.J. et al. (1991)  "Preventing  premature  convergence  in
     GENETIC ALGORITHMs by preventing incest", [ICGA91], 115-122.

     Holland,  J.H.  (1986)  "Escaping  brittleness:  The possibilities of
     general-purpose learning algorithms applied  to  parallel  rule-based
     systems".   In R. Michalski, J. Carbonell, T. Mitchell (eds), Machine
     Learning: An ARTIFICIAL  INTELLIGENCE  Approach.  Los  Altos:  Morgan
     Kaufmann.

     Riolo,   R.L.   (1988)   "CFS-C:  A  package  of  domain  independent
     subroutines for implementing CLASSIFIER SYSTEMs in  arbitrary,  user-
     defined   environments".   Logic  of  computers  group,  Division  of
     computer science and engineering, University of Michigan.

     Grefenstette, J.J. (1989) "A system for learning  control  strategies
     with genetic algorithms", [ICGA89], 183-190.

     De  Jong  K.A.  &  Spears  W. (1991) "Learning concept classification
     rules using genetic algorithms". Proc. 12th IJCAI,  651-656,  Sydney,
     Australia: Morgan Kaufmann.

     Janikow   C.  (1991)  "Inductive  learning  of  decision  rules  from
     attribute-based examples:  A  knowledge-intensive  GENETIC  ALGORITHM
     approach". TR91-030, The University of North Carolina at Chapel Hill,
     Dept. of Computer Science, Chapel Hill, NC.
Go Back Up

Go To Previous

Go To Next