kumulant

UcbTreePosterior

data class UcbTreePosterior(val priorWeight: Double = 1.0, val priorVariance: Double = 1.0) : TreePosterior(source)

UCB-style score: mean + exploration * sqrt(variance / (totalWeights + priorWeight)). The sqrt(.) term is the leaf's standard error of the mean; the prior-weight floor keeps the bound finite at empty leaves.

Constructors

Link copied to clipboard
constructor(priorWeight: Double = 1.0, priorVariance: Double = 1.0)

Properties

Link copied to clipboard

Prior variance used when the leaf has no signal yet.

Link copied to clipboard

Pseudo-count added to the leaf's totalWeights; floor for empty leaves.

Functions

Link copied to clipboard
open override fun evaluate(snapshot: TreeRegressionResult, x: VectorView, rng: Random, exploration: Double = 1.0): Double

Score a query point x under the regression snapshot. exploration controls the posterior-variance scale (Thompson) or the UCB width (LinUcb-style); 0.0 collapses to the point estimate.

UcbTreePosterior

constructor(priorWeight: Double = 1.0, priorVariance: Double = 1.0)(source)

evaluate

open override fun evaluate(snapshot: TreeRegressionResult, x: VectorView, rng: Random, exploration: Double = 1.0): Double(source)

Score a query point x under the regression snapshot. exploration controls the posterior-variance scale (Thompson) or the UCB width (LinUcb-style); 0.0 collapses to the point estimate.

priorVariance

Prior variance used when the leaf has no signal yet.

priorWeight

Pseudo-count added to the leaf's totalWeights; floor for empty leaves.