lasagne.init

Functions to create initializers for parameter variables.

Examples

>>> from lasagne.layers import DenseLayer
>>> from lasagne.init import Constant, GlorotUniform
>>> l1 = DenseLayer((100,20), num_units=50,
...                 W=GlorotUniform('relu'), b=Constant(0.0))

Initializers

Constant([val]) Initialize weights with constant value.
Normal([std, mean]) Sample initial weights from the Gaussian distribution.
Uniform([range, std, mean]) Sample initial weights from the uniform distribution.
Glorot(initializer[, gain, c01b]) Glorot weight initialization.
GlorotNormal([gain, c01b]) Glorot with weights sampled from the Normal distribution.
GlorotUniform([gain, c01b]) Glorot with weights sampled from the Uniform distribution.
He(initializer[, gain, c01b]) He weight initialization.
HeNormal([gain, c01b]) He initializer with weights sampled from the Normal distribution.
HeUniform([gain, c01b]) He initializer with weights sampled from the Uniform distribution.
Orthogonal([gain]) Intialize weights as Orthogonal matrix.
Sparse([sparsity, std]) Initialize weights as sparse matrix.

Detailed description

class lasagne.init.Initializer[source]

Base class for parameter tensor initializers.

The Initializer class represents a weight initializer used to initialize weight parameters in a neural network layer. It should be subclassed when implementing new types of weight initializers.

sample(shape)[source]

Sample should return a theano.tensor of size shape and data type theano.config.floatX.

Parameters:

shape : tuple or int

Integer or tuple specifying the size of the returned matrix.

returns : theano.tensor

Matrix of size shape and dtype theano.config.floatX.

class lasagne.init.Constant(val=0.0)[source]

Initialize weights with constant value.

Parameters:

val : float

Constant value for weights.

class lasagne.init.Normal(std=0.01, mean=0.0)[source]

Sample initial weights from the Gaussian distribution.

Initial weight parameters are sampled from N(mean, std).

Parameters:

std : float

Std of initial parameters.

mean : float

Mean of initial parameters.

class lasagne.init.Uniform(range=0.01, std=None, mean=0.0)[source]

Sample initial weights from the uniform distribution.

Parameters are sampled from U(a, b).

Parameters:

range : float or tuple

When std is None then range determines a, b. If range is a float the weights are sampled from U(-range, range). If range is a tuple the weights are sampled from U(range[0], range[1]).

std : float or None

If std is a float then the weights are sampled from U(mean - np.sqrt(3) * std, mean + np.sqrt(3) * std).

mean : float

see std for description.

class lasagne.init.Glorot(initializer, gain=1.0, c01b=False)[source]

Glorot weight initialization.

This is also known as Xavier initialization [R4].

Parameters:

initializer : lasagne.init.Initializer

Initializer used to sample the weights, must accept std in its constructor to sample from a distribution with a given standard deviation.

gain : float or ‘relu’

Scaling factor for the weights. Set this to 1.0 for linear and sigmoid units, to ‘relu’ or sqrt(2) for rectified linear units, and to sqrt(2/(1+alpha**2)) for leaky rectified linear units with leakiness alpha. Other transfer functions may need different factors.

c01b : bool

For a lasagne.layers.cuda_convnet.Conv2DCCLayer constructed with dimshuffle=False, c01b must be set to True to compute the correct fan-in and fan-out.

See also

GlorotNormal
Shortcut with Gaussian initializer.
GlorotUniform
Shortcut with uniform initializer.

Notes

For a DenseLayer, if gain='relu' and initializer=Uniform, the weights are initialized as

\[\begin{split}a &= \sqrt{\frac{12}{fan_{in}+fan_{out}}}\\ W &\sim U[-a, a]\end{split}\]

If gain=1 and initializer=Normal, the weights are initialized as

\[\begin{split}\sigma &= \sqrt{\frac{2}{fan_{in}+fan_{out}}}\\ W &\sim N(0, \sigma)\end{split}\]

References

[R4](1, 2) Xavier Glorot and Yoshua Bengio (2010): Understanding the difficulty of training deep feedforward neural networks. International conference on artificial intelligence and statistics.
class lasagne.init.GlorotNormal(gain=1.0, c01b=False)[source]

Glorot with weights sampled from the Normal distribution.

See Glorot for a description of the parameters.

class lasagne.init.GlorotUniform(gain=1.0, c01b=False)[source]

Glorot with weights sampled from the Uniform distribution.

See Glorot for a description of the parameters.

class lasagne.init.He(initializer, gain=1.0, c01b=False)[source]

He weight initialization.

Weights are initialized with a standard deviation of \(\sigma = gain \sqrt{\frac{1}{fan_{in}}}\) [R5].

Parameters:

initializer : lasagne.init.Initializer

Initializer used to sample the weights, must accept std in its constructor to sample from a distribution with a given standard deviation.

gain : float or ‘relu’

Scaling factor for the weights. Set this to 1.0 for linear and sigmoid units, to ‘relu’ or sqrt(2) for rectified linear units, and to sqrt(2/(1+alpha**2)) for leaky rectified linear units with leakiness alpha. Other transfer functions may need different factors.

c01b : bool

For a lasagne.layers.cuda_convnet.Conv2DCCLayer constructed with dimshuffle=False, c01b must be set to True to compute the correct fan-in and fan-out.

See also

HeNormal
Shortcut with Gaussian initializer.
HeUniform
Shortcut with uniform initializer.

References

[R5](1, 2) Kaiming He et al. (2015): Delving deep into rectifiers: Surpassing human-level performance on imagenet classification. arXiv preprint arXiv:1502.01852.
class lasagne.init.HeNormal(gain=1.0, c01b=False)[source]

He initializer with weights sampled from the Normal distribution.

See He for a description of the parameters.

class lasagne.init.HeUniform(gain=1.0, c01b=False)[source]

He initializer with weights sampled from the Uniform distribution.

See He for a description of the parameters.

class lasagne.init.Orthogonal(gain=1.0)[source]

Intialize weights as Orthogonal matrix.

Orthogonal matrix initialization [R6]. For n-dimensional shapes where n > 2, the n-1 trailing axes are flattened. For convolutional layers, this corresponds to the fan-in, so this makes the initialization usable for both dense and convolutional layers.

Parameters:

gain : float or ‘relu’

Scaling factor for the weights. Set this to 1.0 for linear and sigmoid units, to ‘relu’ or sqrt(2) for rectified linear units, and to sqrt(2/(1+alpha**2)) for leaky rectified linear units with leakiness alpha. Other transfer functions may need different factors.

References

[R6](1, 2) Saxe, Andrew M., James L. McClelland, and Surya Ganguli. “Exact solutions to the nonlinear dynamics of learning in deep linear neural networks.” arXiv preprint arXiv:1312.6120 (2013).
class lasagne.init.Sparse(sparsity=0.1, std=0.01)[source]

Initialize weights as sparse matrix.

Parameters:

sparsity : float

Exact fraction of non-zero values per column. Larger values give less sparsity.

std : float

Non-zero weights are sampled from N(0, std).