eduzhai > Applied Sciences > Engineering >

Cross entropy as objective function for music generative models

  • Save

... pages left unread,continue reading

Document pages: 6 pages

Abstract: The election of the function to optimize when training a machine learningmodel is very important since this is which lets the model learn. It is nottrivial since there are many options, each for different purposes. In the caseof sequence generation of text, cross entropy is a common option because of itscapability to quantify the predictive behavior of the model. In this paper, wetest the validity of cross entropy for a music generator model with anexperiment that aims to correlate improvements in the loss value with thereduction of randomness and the ability to keep consistent melodies. We alsoanalyze the relationship between these two aspects which respectively relate toshort and long term memory and how they behave and are learned differently.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×