Learning Likelihood Noise

On occasion, it is useful to also learn additional likelihood noise, beyond any known fixed noise, when training a Gaussian process. The LearnYNoise decorator makes this easy.

class vanguard.learning.LearnYNoise(**kwargs)

Bases: Decorator

Learn the likelihood noise.

This decorator passes the appropriate arguments to allow a GPController class to set the likelihood noise as unknown and subsequently learn it.

Example:

>>> @LearnYNoise()
... class NewController(GPController):
...     pass

Parameters:

kwargs (Any)

__init__(**kwargs)

Initialise self.

Parameters:

kwargs (Any) – Keyword arguments passed to Decorator.

property safe_updates: dict[type, set[str]]

Get a dictionary (class -> set[names]) of overrides/new methods that we consider “safe”.