My guess is that neural networks do not work very well in noisy environments, i.e. the lower the signal-to-noise ratio, the worse the result of a neural network, if compared to other statistical modeling tools. Thus, for example, neural networks are good at predicting credit cards frauds, but they get much less exciting results when trying to predict financial markets (very noisy, at least in the short term).

Any theoretical result and/or empirical evidence on that?

  • 1
    $\begingroup$ This is an overly broad question. Can you be more specific? $\endgroup$ – Arun Jose Oct 28 '16 at 9:46
  • 1
    $\begingroup$ IMHO, this is not too broad. There should be some generic knowledge about this, after all. $\endgroup$ – Richard Hardy Oct 28 '16 at 17:42
  • $\begingroup$ Why would this be the case? Why would other ML approaches be any better in situations where there is substantial idiosyncratic error? $\endgroup$ – generic_user Oct 28 '16 at 19:33
  • $\begingroup$ If you overfit a network it will try to follow the noise as well as the signal, but this can happen for spline and polynomial fits too. There are techniques to avoid overfitting. $\endgroup$ – KAE Oct 2 '18 at 19:38

Your Answer

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

Browse other questions tagged or ask your own question.