1
$\begingroup$

This is what I am trying to accomplish:

Predict Y based on X,that is Y ~ X:

1.X = {n, x_1, x_2, x_3, ...}

  • a vector of factors
  • size unchanged: X.size() = DIM_X

2.Y = {}

  • also a vector
  • size changing according to n in X: Y.size() = 4 * n

So, can I do this with NNs ? is there a NN-Model I can refer to ? Or Is there other approaches can solve this ？

Edit:

Sample data:

1 0.1 109 123.7 0 0 512 512
2 0.8 201 88.7 384 216 384 246 0 0 1024 768

In the above, first 4 elements each row represents X, what's left is Y.

Appreciate your help.

Improve this question
$\endgroup$
8
  • 1
    $\begingroup$ Ignore my first comment (if you saw it). Your question is not very clear though. Perhaps add a few simplified data examples. Probably a RNN model can do what you want. $\endgroup$
    – Neil Slater
    Nov 1, 2016 at 10:25
  • 1
    $\begingroup$ I agree with some examples and some more info about n, is n bounded and if so what is the cardinality? $\endgroup$
    – Jan van der Vegt
    Nov 1, 2016 at 10:42
  • $\begingroup$ Thanks for pointing out, see my edits above, is it more clear? $\endgroup$
    – xtluo
    Nov 1, 2016 at 10:51
  • 1
    $\begingroup$ Some info about n? If that is always 10 or less is a very different problem than if it can grow up to 2000 $\endgroup$
    – Jan van der Vegt
    Nov 1, 2016 at 11:19
  • 1
    $\begingroup$ In your sample data, are the first four features of Y representing the same things for n = 1 and n = 2? (I.e., are they coming from the same probability distribution?) $\endgroup$
    – J. O'Brien Antognini
    Nov 1, 2016 at 17:38

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.