MultilayerPerceptron Class Reference

Inheritance diagram for MultilayerPerceptron:

Public Member Functions
	__init__ (self, layers, name='mlp')
	from_list (cls, layers)
	initialize (self)
	__call__ (self, x)
	variables_to_writer (self, step, writer)

Public Attributes
	layers = layers
	layer objs
list	w = None
	weights
list	b = None
	biases
bool	is_initialized = False
	collect all mlp parameters

Protected Member Functions
	_collect_weights_and_biases (self)

Detailed Description

multilayer perceptron class.

Definition at line 118 of file tensorflow_dnn_model.py.

Constructor & Destructor Documentation

__init__()

__init__	(		self,
			layers,
			name = 'mlp' )

initialization

Definition at line 123 of file tensorflow_dnn_model.py.

    def __init__(self, layers, name='mlp'):
        """
        initialization
        """
        super().__init__(name=name)
 
        
        self.layers = layers
 
        
        self.w = None
 
        
        self.b = None
 
        
        self.is_initialized = False
        self.initialize()
 

Member Function Documentation

__call__()

__call__	(		self,
			x )

Run the events through all the layers

Definition at line 181 of file tensorflow_dnn_model.py.

    def __call__(self, x):
        """
        Run the events through all the layers
        """
        for layer in self.layers:
            x = layer(x)
        return x
 

_collect_weights_and_biases()

_collect_weights_and_biases ( self )

protected

collect tunable parameters

Definition at line 154 of file tensorflow_dnn_model.py.

    def _collect_weights_and_biases(self):
        """
        collect tunable parameters
        """
        self.w = []
        self.b = []
        for layer in self.layers:
            self.w.append(layer.w)
            self.b.append(layer.b)
 

from_list()

from_list	(		cls,
			layers )

define layers from list

Definition at line 143 of file tensorflow_dnn_model.py.

    def from_list(cls, layers):
        """
        define layers from list
        """
        layer_obj = []
        for layer in layers:
            layer_obj.append(Layer(*layer))
 
        mlp = cls(layer_obj)
        return mlp
 

initialize()

initialize

(

self

)

initialize. Checks that the layer dimensions align.

Definition at line 164 of file tensorflow_dnn_model.py.

    def initialize(self):
        """
        initialize. Checks that the layer dimensions align.
        """
        if self.is_initialized:
            raise RuntimeError
 
        # check shape
        for _idx in range(len(self.layers) - 1):
            assert self.layers[_idx].shape[1] == self.layers[_idx + 1].shape[0]
 
        self._collect_weights_and_biases()  # TODO - remove?
 
        self.is_initialized = True
        return
 

variables_to_writer()

variables_to_writer	(		self,
			step,
			writer )

passes all the MLP variables to the tf.summary writer

Definition at line 189 of file tensorflow_dnn_model.py.

    def variables_to_writer(self, step, writer):
        """
        passes all the MLP variables to the tf.summary writer
        """
        for layer in self.layers:
            layer.all_to_summary(step, writer)
        return
 
 

Member Data Documentation

b

list b = None

biases

Definition at line 136 of file tensorflow_dnn_model.py.

is_initialized

bool is_initialized = False

collect all mlp parameters

Definition at line 139 of file tensorflow_dnn_model.py.

layers

layers = layers

layer objs

Definition at line 130 of file tensorflow_dnn_model.py.

w

list w = None

weights

Definition at line 133 of file tensorflow_dnn_model.py.

---

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

• analysis/scripts/dft/tensorflow_dnn_model.py