Auto update of hyper parameters

Issue #15 resolved
Balaji Pokuri created an issue

Current implementation enables manual change of the hyper-parameters. However there are methods using maximum likelihood that help to change them to suit the exact problem landscape. This will reduce manual tweaking of bounds and re-maps

Official response

  • Balaji Pokuri reporter
    • changed status to open

    The current implementation of estimating MLE does not support external kernel functions. The kernel functions may need to be re-written as a class with a set_parameter function, and the BayesOpt should take in the class instance and not just a function/string.

Comments (6)

  1. Balaji Pokuri reporter

    To be tested... call the function estimate_best_kernel_parameters() for applying the best kernel parameters through log-likelihood calculation. Added in 311da02

  2. Balaji Pokuri reporter
    • changed status to open

    The current implementation of estimating MLE does not support external kernel functions. The kernel functions may need to be re-written as a class with a set_parameter function, and the BayesOpt should take in the class instance and not just a function/string.

  3. Log in to comment