Geek Answers Handbook

What is the purpose of using a genetic algorithm in the study of ANN

I learned the basics of learning ANN with a genetic algorithm. I found out that there are basically 2 things you can do:

  • Use GA to design the network structure (determine if there should be a border between two neurons or not). I suggest that we can only use a certain amount of neuron-neural connections.
  • Use GA to calculate optimal weights.

I also found out that GA only makes sense in the case of irregular networks. If the network consists of layers, she suggested using back propagation as quickly as possible.

If backpropagation is faster and requires a layered network, why would I want to choose GA for training or network design?

+4
source share
3 answers

Use GA to design the network structure (determine if there should be an edge between two neurons or not).

In general, you seem to be talking about direct-connect networks, presumably MLP.

The number of neurons and layers, as well as the connections between neurons, are associated with this network structure. Usually they are completely connected, so each neuron in layer n is connected to each layer n + 1. The training method will determine partial connectivity by training some weights to zero or very small numbers.

ANN , . . , .

, GA , .

GA .

GA . " " . . " " .

MLP a GA , Back Propagation.

, GA , , , GA, BP.

. . GA, BP, . , . /, . , , , , .

, GA . , .

, BP. , , , GA.

. GA . , , , .

, , BP, ANN , MLP!

, , GA ?

FF, BP . . A GA , , ,....

, , , Back Propagation , . GA - , , "" . /.

; , , , , . GA; , , , ,... .

+1

( ). , , , . GA , .

. . , , . .

, , ANN.

+1

, ; GA, , BP .

0

More articles:


All Articles