mxnet.gluon.probability.transformation

Transformation classes.

Classes

AbsTransform()
AffineTransform(loc, scale[, event_dim])	Perform pointwise affine transform: y = loc + scale * x.
ComposeTransform(parts)	Composes multiple transforms in a chain.
ExpTransform()	Perform the exponential transform: y = exp{x}.
PowerTransform(exponent)	Perform pointwise power transform: y = pow(x, exponent).
SigmoidTransform()	Perform pointwise sigmoid transform: y = 1 / (1 + exp(-x)).
SoftmaxTransform()
TransformBlock(*args, **kwargs)	Transform with learnable parameters should inherit from this class rather than Transformation.
Transformation()	Abstract class for implementing invertible transformation with computable log det jacobians