Home Original page

ANN: pygene - genetic algorithms package

Erik Max Francis max at alcyone.com
Thu Dec 8 23:23:03 EST 2005 
malv wrote:

> How is your package different from a nn package? Is this an addon for
> genetic programming or does it include the standard nn components as
> well, such as backprop etc?
> Not being very familiar with genetic programming, forgive me my naive
> question, I could not immediately find the answer.

No one answered your question, so I will:  Neural networks and genetic 
algorithms/programming are not closely related.  A neural network is a 
model comprising a network of artificial neurons, and signals propagate 
from inputs, through the neurons of the network (which typically 
consists of several layers), and finally to the outputs.  Algorithms 
exist to allow neural networks to be trained, so that given a set of 
inputs and desired outputs, one can iteratively come up with a neural 
network (of sufficient complexity) that satisfies that criterion. 
Neural networks can be used for things like pattern recognition and the 
like.

Genetic algorithms/programming is an approach whereby you can start with 
a population of random programs, apply artificial selection by keeping 
those that do better at a task than others and genetic modification like 
mutation and crossover (simulation sexual reproduction) to create 
diversity, ultimately resulting (hopefully) in programs that are adept 
at the task.

They could be used in conjunction with one another (for instance, a 
genetic algorithm where the program is represented by an array of neural 
network weights), but beyond intermixing ideas they really aren't related.

-- 
Erik Max Francis && max at alcyone.com && http://www.alcyone.com/max/
San Jose, CA, USA && 37 20 N 121 53 W && AIM erikmaxfrancis
   I never could have predicted / That I'd feel this way
   -- India Arie
More information about the Python-list mailing list