OpenANN::Agent Class Reference

A (learning) agent in a reinforcement learning problem. More...

#include <Agent.h>

Inheritance diagram for OpenANN::Agent:

Public Member Functions
virtual	~Agent ()
virtual void	abandoneIn (Environment &environment)=0
	Abandon an agent in an environment. More...
virtual void	chooseAction ()=0
	Choose an action and execute it in the environment. More...
virtual void	chooseOptimalAction ()=0
	Choose an action and execute it in the environment. More...

Detailed Description

A (learning) agent in a reinforcement learning problem.

Constructor & Destructor Documentation

virtual OpenANN::Agent::~Agent ( )

inlinevirtual

Member Function Documentation

virtual void OpenANN::Agent::abandoneIn ( Environment &  environment)

pure virtual

Abandon an agent in an environment.

Parameters

environment

reinforcement learning environment

Implemented in NeuroEvolutionAgent, and OpenANN::RandomAgent.

virtual void OpenANN::Agent::chooseAction ( )

pure virtual

Choose an action and execute it in the environment.

This action might not be optimal, i.e. the agent is allowed to explore.

Implemented in NeuroEvolutionAgent, and OpenANN::RandomAgent.

virtual void OpenANN::Agent::chooseOptimalAction ( )

pure virtual

Choose an action and execute it in the environment.

The action must be optimal, i.e. the agent must exploit the learned policy.

Implemented in NeuroEvolutionAgent, and OpenANN::RandomAgent.

---

The documentation for this class was generated from the following file:

• OpenANN/rl/Agent.h