kumulant/com.eignex.kumulant.bandit.univariate/EpsilonGreedy

EpsilonGreedy

class EpsilonGreedy(val epsilon: Double = 0.1, priorMean: Double = 0.0, priorWeight: Double = 0.02, priorSquaredDeviations: Double = 0.02) : BanditPolicy<WeightedVarianceResult> (source)

Epsilon-greedy: with probability epsilon pick a uniformly random arm (explore), otherwise pick the arm with the highest mean (exploit). The simplest exploration scheme that actually works; no math machinery, tune one knob.

Sensitive to the epsilon value: too low and you under-explore (regret scales linearly in horizon for the wrong arm); too high and you waste pulls on known-bad arms. Typical good values are 0.05–0.2. For automatic tuning use EpsilonDecreasing, which anneals epsilon toward zero as samples accumulate.

Constructors

EpsilonGreedy

constructor(epsilon: Double = 0.1, priorMean: Double = 0.0, priorWeight: Double = 0.02, priorSquaredDeviations: Double = 0.02)

Properties

arm

open override val arm: NormalArm

Per-arm cumulator spec; determines the prior pseudo-counts, value encoding, and result shape that evaluate consumes.

epsilon

val epsilon: Double

Probability of exploring uniformly instead of exploiting the best mean.

Functions

addArm

open fun addArm(snapshot: WeightedVarianceResult)

Hook called when a new arm joins the population. Lets stateful policies fold the new arm's snapshot into their global counters (UCB's total-samples, UCB1Normal's arm count). Default no-op.

createArm

open fun createArm(): SeriesStat<WeightedVarianceResult>

Allocate a fresh per-arm accumulator from the arm spec. Default delegates to arm.createStat(); override only if the policy needs a non-standard variant.

evaluate

open override fun evaluate(snapshot: WeightedVarianceResult, step: Long, rng: Random): Double

Score an arm given its current snapshot. Higher scores are preferred by the bandit. step is the global update count (for time-dependent exploration schedules); rng is the bandit's shared com.eignex.kumulant.bandit.Bandit.random (consumed by sampling policies).

removeArm

open fun removeArm(snapshot: WeightedVarianceResult)

Hook called when an arm leaves the population. Inverse of addArm; lets stateful policies remove the departing arm's contribution from their global counters. Default no-op.

update

open fun update(stat: SeriesStat<WeightedVarianceResult>, value: Double, weight: Double = 1.0)

Fold an observed reward value (with optional weight) into the per-arm stat. Default applies arm.encode first; policies with global counters (UCB families) override to update their counter alongside the stat update.