prevision_quantum_nn.applications.reinforcement_learning_application

Reinforcement Learning module provides with the class for reinforcement learning applications

Module Contents

Classes

ReinforcementLearningApplication

Reinforcement Learning application.

class prevision_quantum_nn.applications.reinforcement_learning_application.ReinforcementLearningApplication(prefix='qnn', preprocessing_params=None, model_params=None, postprocessing_params=None, rl_learner_type='quantum')

Bases: prevision_quantum_nn.applications.application.Application

Reinforcement Learning application.

value_function

(str):type of the value function, can be either state or state_action

environment

environment to be solved

Type:gym.env or custom environment

num_acitons

number of possible actions for the agent

Type:int

build(self)

Build the Application (learner, policy).

solve(self, environment, tqdm_verbose=True, render=False)

Solves the problem given a certain environment.

Parameters:

• environment (gym like env) – environment of RL app
• tqdm_verbose (bool) – activates tqdm (optionnal)
• render (bool) – activates rendering (optionnal)

preprocess_state(self, state)

Preprocess the state.

Parameters:state (int or state) – state, discrete or continuous Returns:

np.array

preprocessed state

Return type:state