keraflow.backend.theano_backend.TheanoBackend Class Reference

Inheritance diagram for keraflow.backend.theano_backend.TheanoBackend:

Public Member Functions
def	variable
	Instantiate a tensor variable.
def	placeholder
	Instantiate an input data placeholder variable.
def	shape
	Returns the symbolic shape of a tensor.
def	eval
	Evaluates the value of a tensor. More...
def	switch
	condition: scalar tensor.
def	zeros
	Instantiate an all-zeros tensor variable.
def	ones
	Instantiate an all-ones tensor variable.
def	eye
	Instantiate an identity matrix.
def	zeros_like
	Instantiates an all-zeros tensor of the same shape as another tensor.
def	ones_like
	Instantiates an all-ones tensor of the same shape as another tensor.
def	dot
	numpy.dot on tensors
def	gather
	Retrieves the vectors of indices indices in the 2D tensor reference. More...
def	prod
	Multiply the values in a tensor, alongside the specified axis.
def	any
	Bitwise reduction (logical OR).
def	transpose
	Transpose dimensions. More...
def	repeat
	Repeat the elements of a tensor along an axis, like np.repeat. More...
def	expand_dims
	Add a 1-sized dimension at index "axis".
def	squeeze
	Remove a 1-dimension from the tensor at index "axis".
def	dropout
	Sets entries in x to zero at random, while scaling the entire tensor. More...
def	rnn
	Iterates over the time dimension of a tensor. More...

Member Function Documentation

def keraflow.backend.theano_backend.TheanoBackend.dropout	(		self,
			x,
			drop_rate,
			noise_shape = None
	)

Sets entries in x to zero at random, while scaling the entire tensor.

Parameters

x	tensor
drop_rate	fraction of the entries in the tensor that will be set to 0.
noise_shape	shape for randomly generated keep/drop flags, must be broadcastable to the shape of x

def keraflow.backend.theano_backend.TheanoBackend.eval	(		self,
			x
	)

Evaluates the value of a tensor.

Returns a Numpy array.

def keraflow.backend.theano_backend.TheanoBackend.gather	(		self,
			reference,
			indices
	)

Retrieves the vectors of indices indices in the 2D tensor reference.

Parameters

reference	a 2D tensor.
indices	2D int tensor or list.

Returns

3D tensor.

def keraflow.backend.theano_backend.TheanoBackend.repeat	(		self,
			x,
			rep,
			axis
	)

Repeat the elements of a tensor along an axis, like np.repeat.

If x has shape (s1, s2, s3) and axis=1, the output will have shape (s1, s2 * rep, s3).

def keraflow.backend.theano_backend.TheanoBackend.rnn	(		self,
			step_function,
			inputs,
			initial_states,
			go_backwards = False,
			mask = None,
			unroll = False,
			input_length = None
	)

Iterates over the time dimension of a tensor.

Arguments

inputs: tensor of temporal data of shape (samples, time, ...) (at least 3D). step_function: Parameters: input: tensor with shape (samples, ...) (no time dimension), representing input for the batch of samples at a certain time step. states: list of tensors. Returns: output: tensor with shape (samples, ...) (no time dimension), new_states: list of tensors, same length and shapes as 'states'. initial_states: tensor with shape (samples, ...) (no time dimension), containing the initial values for the states used in the step function. go_backwards: boolean. If True, do the iteration over the time dimension in reverse order. mask: binary tensor with shape (samples, time), with a zero for every element that is masked. unroll: whether to unroll the RNN or to use a symbolic loop (scan). input_length: must be specified if using unroll.

Returns

A tuple (last_output, outputs, new_states). last_output: the latest output of the rnn, of shape (samples, ...) outputs: tensor with shape (samples, time, ...) where each entry outputs[s, t] is the output of the step function at time t for sample s. new_states: list of tensors, latest states returned by the step function, of shape (samples, ...).

def keraflow.backend.theano_backend.TheanoBackend.transpose	(		self,
			x,
			dims
	)

Transpose dimensions.

dims should be a tuple or list of dimension indices, e.g. [0, 2, 1].