pashehnet.sensors.transforms.dropout#

Module Contents#

Classes#

DropoutTransform

Signal transform that provides the following features:

API#

class pashehnet.sensors.transforms.dropout.DropoutTransform(prob=0.01, value=0.0, duration=1, duration_range=None, rng=None)#

Bases: pashehnet.sensors.transforms.base.SensorTransformBase

Signal transform that provides the following features:

  • Probability of dropout starting (uniform distro)

  • Value to provide during dropout

  • Duration of dropout, one of - Time (constant or min/max variable from uniform distro) - Sequence count (constant or min/max variable from uniform distro)

Initialization

CTOR for class

Parameters:

  • prob – Probability of dropout [0.0, 1.0]

  • value – Value to use when dropout of signal occurs

  • duration – Sample count in dropout

  • duration_range – Tuple of min/max dropout counts

  • rng – NumPy random number generator to use; defaults to numpy.random.default_rng

__slots__ = ()#
transform(value)#

Apply transform, calculating if a dropout is occurring and returning appropriate value

Parameters:

value – Value to apply transform to

Returns:

Transformed value