Tell me more ×
Programmers Stack Exchange is a question and answer site for professional programmers interested in conceptual questions about software development. It's 100% free, no registration required.
up vote -2 down vote favorite

Backpropagation is perhaps the most famous method for training neural networks. What are some other alternative methods for training neural networks and how do they compare to backpropagation?

share|improve this question
en.wikipedia.org/wiki/…: "Evolutionary methods, gene expression programming, simulated annealing, expectation-maximization, non-parametric methods and particle swarm optimization are some commonly used methods for training neural networks." – Robert Harvey Nov 29 '12 at 22:00
What can I add to make this question more clear? I've already received two answers (one of which was a comment) from people who didn't have any problem understanding the question. – Buttons840 Nov 30 '12 at 17:31
Your question is already more than adequately answered elsewhere. Spend some time doing some of your own research (see the Wikipedia article I linked), and come back here with a more specific, focused question if you need clarification on some aspect of your research. From the faq: "Your questions should be reasonably scoped. If you can imagine an entire book that answers your question, you’re asking too much." – Robert Harvey Nov 30 '12 at 17:47
@RobertHarvey: It says it's closed because "It's difficult to tell what is being asked here." That isn't true. It's very easy to tell what's being asked here. Closing it may be useful, but the closure message is false. – Michael Shaw Dec 1 '12 at 3:59
   
@MichaelShaw: The phrases in the close message are intended to be as all-encompassing as possible. Some questions are closed because they are ambiguous, others because they are vague, incomplete or rhetorical. This one was closed because it was overly broad. It's not always difficult to tell what is being asked, but it can still be "Not a Real Question" nevertheless. – Robert Harvey Dec 1 '12 at 4:39
show 5 more comments

closed as not a real question by Robert Harvey, Walter, Mason Wheeler, ElYusubov, gnat Nov 30 '12 at 8:49

It's difficult to tell what is being asked here. This question is ambiguous, vague, incomplete, overly broad, or rhetorical and cannot be reasonably answered in its current form. For help clarifying this question so that it can be reopened, see the FAQ.

1 Answer

up vote 1 down vote

Reinforcement learning algorithms such as Neuroevolution are an alternative. They are probably the most well known way of training neural networks for problems that aren't as well defined (for example, playing certain games). However I have found in practice that neural networks by themselves are virtually always a suboptimal way of tackling these problems.

If you're interested in the area, start by checking out NEAT, it's the easiest, best method I've found for implementing good, workable neuroevolution.

share|improve this answer

Not the answer you're looking for? Browse other questions tagged or ask your own question.