Neural Network Programming with TensorFlow

Class initialization

In DBN initialization, the Model class's initialization method __init__(self, name) is called. The Model class references the following:

• Input data: self.input_data
• Input labels: self.input_labels
• Cost: self.cost
• Number of nodes in final layer: self.layer_nodes
• TensorFlow session: self.tf_session
• TensorFlow graph: self.tf_graph= tf.graph

class Model(object):    """Class representing an abstract Model."""    def __init__(self, name):        """Constructor.        :param name: name of the model, used as filename.            string, default 'dae'        """        self.name = name        self.model_path = os.path.join(Config().models_dir, self.name)        self.input_data = None        self.input_labels = None>        self.keep_prob = None        self.layer_nodes = []  # list of layers of the final ...