I'm looking to design a neural network that can predict which runner wins in a sports game, where the amount of runners varies between 2-10. In each case, specific data about the individual runners would be fed into the neural network.

What design would be most advantageous for such a neural network?

Essentially this is a ranking problem where the amount of inputs and outputs are variable.

  • $\begingroup$ Downvoter reveal yourself, what's the problem with that question? $\endgroup$ – Nickpick Oct 16 '17 at 17:26
  • $\begingroup$ +1 Hi @nickpick, can you explain a little bit what you mean by "specific data about the individual runners would be fed into the neural network." ? Can you give an example or two? Ps, I'm not the down-voter(s) :) $\endgroup$ – Tshilidzi Mudau Oct 17 '17 at 12:36
  • $\begingroup$ For example, weight, average speed in previous races, heigth, nationality etc $\endgroup$ – Nickpick Oct 17 '17 at 12:37
  • 1
    $\begingroup$ @nickpick yeah, it's a problem on this site - ai.meta.stackexchange.com/questions/1313/… $\endgroup$ – Maxim Oct 17 '17 at 13:02
  • 1
    $\begingroup$ Welcome to AI. Glad this question garnered several upvotes! $\endgroup$ – DukeZhou Oct 18 '17 at 20:17

The best option in your case would probably be zero padding or padding up. This is simply zeroing out inputs for cases in which there is no data. It's done a lot on the borders of images for CNNs.

That or you could just use a RNN which can handle your variable length inputs with ease.

  • $\begingroup$ But can the amount of output neurones vary in accordance with how many inputs are made in a RNN? This would be necessary to rank the individual runners. $\endgroup$ – Nickpick Oct 18 '17 at 22:07
  • $\begingroup$ Yes, RNNs can handle variable length outputs as well. $\endgroup$ – hisairnessag3 Oct 18 '17 at 22:13
  • $\begingroup$ Are there any examples or documentations? I’m using keras $\endgroup$ – Nickpick Oct 18 '17 at 23:41
  • 1
    $\begingroup$ Keras Docs on RNNs: keras.io/layers/recurrent Github issue on this for Keras: github.com/fchollet/keras/issues/40 $\endgroup$ – hisairnessag3 Oct 18 '17 at 23:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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