ContextualEpsilonGreedyPolicy: Policy: ContextualEpsilonGreedyPolicy with unique linear...

In contextual: Simulation and Analysis of Contextual Multi-Armed Bandit Policies

Description

Policy: ContextualEpsilonGreedyPolicy with unique linear models

Usage

policy <- ContextualEpsilonGreedyPolicy(epsilon = 0.1)

Arguments

epsilon

double, a positive real value R+

Parameters

A

d*d identity matrix

b

a zero vector of length d

Methods

new(epsilon = 0.1)

Generates a new ContextualEpsilonGreedyPolicy object. Arguments are defined in the Argument section above.

set_parameters()

each policy needs to assign the parameters it wants to keep track of to list self$theta_to_arms that has to be defined in set_parameters()'s body. The parameters defined here can later be accessed by arm index in the following way: theta[[index_of_arm]]$parameter_name

get_action(context)

here, a policy decides which arm to choose, based on the current values of its parameters and, potentially, the current context.

set_reward(reward, context)

in set_reward(reward, context), a policy updates its parameter values based on the reward received, and, potentially, the current context.

See Also

Core contextual classes: Bandit, Policy, Simulator, Agent, History, Plot

Bandit subclass examples: BasicBernoulliBandit, ContextualLogitBandit, OfflineReplayEvaluatorBandit

Policy subclass examples: EpsilonGreedyPolicy, ContextualLinTSPolicy

contextual documentation built on July 26, 2020, 1:06 a.m.