prevision_quantum_nn.applications.reinforcement_learning.deep_q_learning

Deep Q Learning module

Module Contents

Classes

DeepQLearner	DeepQLearner.
DeepFullyConnectedLearner	Deep Fully Connected Learner.

class prevision_quantum_nn.applications.reinforcement_learning.deep_q_learning.DeepQLearner(params=None)

Bases: prevision_quantum_nn.applications.reinforcement_learning.base_learner.BaseLearner

DeepQLearner.

Base class for further implementations of Deep Q Learners

params

contains the parameters of the model

Type:dictionary

input_size

the size of the state space

Type:int

model

the model itself

Type:tf.keras.model

optimizer_name

the name of the optimizer, can be adam for example

Type:str

fit(self, x_train, y_train)

Fit the model.

Parameters:

• x_train (numpy array) – contains the features of the observations
• y_train (numpy array) – contains the targets of the observations

class prevision_quantum_nn.applications.reinforcement_learning.deep_q_learning.DeepFullyConnectedLearner(params=None)

Bases: prevision_quantum_nn.applications.reinforcement_learning.deep_q_learning.DeepQLearner

Deep Fully Connected Learner.

This deep model is based on a fully connected neural network structure.

build(self)

builds the keras model given a fixed strucutre

forward(self, state)

Forwards a state through the neural network.

Parameters:state (numpy array) – state at which the model needs to predict the Q-value