Discussion :

[Djug]

Bonjour,

La rédaction DVP a lu pour vous le livre "A Field Guide to Genetic Programming" de Riccardo Poli, William B. Langdon et Nicholas Freitag McPhee

Genetic programming (GP) is a systematic, domain-independent method for getting computers to solve problems automatically starting from a high-level statement of what needs to be done. Using ideas from natural evolution, GP starts from an ooze of random computer programs, and progressively refines them through processes of mutation and sexual recombination, until high-fitness solutions emerge. All this without the user having to know or specify the form or structure of solutions in advance. GP has generated a plethora of human-competitive results and applications, including novel scientific discoveries and patentable inventions. This unique overview of this exciting technique is written by three of the most active scientists in GP.

L'avez-vous lu ? Comptez-vous le lire bientôt ?

Quel est votre avis ?

Exprimez-vous !! Votre avis nous intéresse...

[rmaker]

L'avez-vous lu ? Comptez-vous le lire bientôt ?

Quel est votre avis ?

Exprimez-vous !! Votre avis nous intéresse...

Lu et beaucoup aimé. Il pose de très bonnes questions (en peu de pages), notamment le code généré inutile (çà a un nom, mais je ne m'en rappelle plus). Il commence par reprendre rapidement ce que sont les algorithmes génétiques dans le cadre général. Par rapport au livre "Linear Genetic Programming", il est vraiment beaucoup plus facile d'accès et lisible.

Bref, je recommande.

[Franck Dernoncourt]

2012 - Free Book - Moshe Sipper - Evolved to Win
http://algo.developpez.com/livres/#L9781470972837

3.2.6 Explicitly defined introns

In natural living systems not all DNA has a phenotypic effect. This non-coding DNA,
sometimes referred to as junk DNA, is prevalent in virtually all eukaryotic genomes.
In GP, so-called introns are areas of code that do not affect survival and reproduction
(usually this can be replaced with “do not affect fitness”). In the context of tree-based
GP the term “areas of code” applies to subtrees.


Introns occur naturally in GP, provided that the function and terminal sets allow
for it. As bloat progresses, the number of nodes that are part of introns tends to
increase. Luke [117] distinguished two types of subtrees that are sometimes referred
to as introns in the literature:
• Unoptimized code: Areas of code that can be trivially simplified without modifying
the individual’s functionality, but not replaced with just anything.
• Inviable code: Subtrees that cannot be replaced by anything that can possibly
change the individual’s functionality.


Luke focused on inviable introns and we will do the same because unoptimized
code seems to cast too wide a net and wander too far from the original meaning of
the term “intron” in biology. We also make another distinction between two types of
inviable-code introns:
• Live-code introns: Subtrees that cannot be replaced by anything that can possibly
change the individual’s functionality, but may still generate code that will run
at some point.
• Dead-code introns: Subtrees whose code is never run.

[Franck Dernoncourt]

Quick biology refresher:

Genes in DNA are interrupted by introns that do not code for proteins but often play an important role in regulation. mRNA has these introns stripped away and only contains exons , i.e. regions that are expressed.