SuccessiveRejectsPolicy.hpp

Go to the documentation of this file.

#ifndef SUCCESSIVE_REJECTS_POLICY_HEADER_FILE
#define SUCCESSIVE_REJECTS_POLICY_HEADER_FILE
 
#include <AIToolbox/Types.hpp>
#include <AIToolbox/Bandit/Experience.hpp>
#include <AIToolbox/Bandit/Policies/PolicyInterface.hpp>
 
namespace AIToolbox::Bandit {
    class SuccessiveRejectsPolicy : public PolicyInterface {
        public:
            SuccessiveRejectsPolicy(const Experience & experience, unsigned budget);
 
            virtual size_t sampleAction() const override;
 
            void stepUpdateQ();
 
            bool canRecommendAction() const;
 
            size_t recommendAction() const;
 
            size_t getCurrentPhase() const;
 
            size_t getCurrentNk() const;
 
            size_t getPreviousNk() const;
 
            virtual double getActionProbability(const size_t & a) const override;
 
            virtual Vector getPolicy() const override;
 
            const Experience & getExperience() const;
 
        private:
            void updateNks();
 
            const Experience & exp_;
            unsigned budget_;
 
            unsigned currentPhase_;
            size_t currentActionId_;
            unsigned currentArmPulls_;
 
            unsigned nKOld_, nKNew_;
            double logBarK_;
            std::vector<size_t> availableActions_;
    };
}
 
#endif