EpsilonPolicy.hpp

Go to the documentation of this file.

#ifndef AI_TOOLBOX_FACTORED_BANDIT_EPSILON_POLICY_HEADER_FILE
#define AI_TOOLBOX_FACTORED_BANDIT_EPSILON_POLICY_HEADER_FILE
 
#include <AIToolbox/Factored/Bandit/Policies/PolicyInterface.hpp>
#include <AIToolbox/EpsilonPolicyInterface.hpp>
 
namespace AIToolbox::Factored::Bandit {
    class EpsilonPolicy : public PolicyInterface, public EpsilonPolicyInterface<void, void, Action> {
        public:
            using EpsilonBase = EpsilonPolicyInterface<void, void, Action>;
 
            EpsilonPolicy(const PolicyInterface & p, double epsilon = 0.1);
 
        protected:
            virtual Action sampleRandomAction() const override;
 
            virtual double getRandomActionProbability() const override;
 
            // Used to sampled random actions
            mutable std::vector<std::uniform_int_distribution<size_t>> randomDistribution_;
    };
}
 
#endif