MAUCEPolicy.hpp

Go to the documentation of this file.

#ifndef AI_TOOLBOX_FACTORED_BANDIT_MAUCE_POLICY_HEADER_FILE
#define AI_TOOLBOX_FACTORED_BANDIT_MAUCE_POLICY_HEADER_FILE
 
#include <AIToolbox/Factored/Bandit/Types.hpp>
#include <AIToolbox/Factored/Utils/FilterMap.hpp>
#include <AIToolbox/Factored/Bandit/Algorithms/Utils/UCVE.hpp>
#include <AIToolbox/Factored/Bandit/Experience.hpp>
#include <AIToolbox/Factored/Bandit/Policies/PolicyInterface.hpp>
 
namespace AIToolbox::Factored::Bandit {
    class MAUCEPolicy : public PolicyInterface {
        public:
            MAUCEPolicy(const Experience & exp, std::vector<double> ranges);
 
            virtual Action sampleAction() const override;
 
            virtual double getActionProbability(const Action & a) const override;
 
            const Experience & getExperience() const;
 
        private:
            const Experience & exp_;
            std::vector<double> rangesSquared_;
            double logA_;
    };
}
 
#endif