KSLCore/ksl.simopt.solvers.algorithms.bo/MleHyperparameterFitter

MleHyperparameterFitter

class MleHyperparameterFitter(numStarts: Int = defaultNumStarts, lengthScaleLowFactor: Double = 0.05, lengthScaleHighFactor: Double = 2.0) : HyperparameterFitterIfc(source)

A maximum-likelihood hyperparameter fitter: it performs a random multi-start search over the log of the length scales and signal variance, refitting the GP at each candidate and keeping the hyperparameters with the highest log marginal likelihood. Length-scale candidates are drawn log-uniformly between lengthScaleLowFactor and lengthScaleHighFactor times each input range; signal-variance candidates are drawn log-uniformly around the data variance.

Parameters

numStarts

the number of random restarts. Must be >= 1. Defaults to defaultNumStarts.

lengthScaleLowFactor

the lower bound factor on length scales. Must be > 0.

lengthScaleHighFactor

the upper bound factor on length scales. Must be >lengthScaleLowFactor.

Constructors

MleHyperparameterFitter

constructor(numStarts: Int = defaultNumStarts, lengthScaleLowFactor: Double = 0.05, lengthScaleHighFactor: Double = 2.0)

Types

object Companion

Properties

lengthScaleHighFactor

var lengthScaleHighFactor: Double

lengthScaleLowFactor

var lengthScaleLowFactor: Double

var numStarts: Int

Functions

open override fun fit(model: GaussianProcessModel, points: List<DoubleArray>, means: DoubleArray, noiseVars: DoubleArray, rnStream: RNStreamIfc)

Sets the kernel hyperparameters of model given the training data. Randomness (if any) is drawn through rnStream.

open override fun toString(): String