ezmsg.learn.process.sgd#

Functions

sgd_decoder(alpha=1.5e-05, eta0=1e-07, loss='squared_hinge', label_weights=None, settings_path=None)[source]#

Passive Aggressive Classifier Online Passive-Aggressive Algorithms <http://jmlr.csail.mit.edu/papers/volume7/crammer06a/crammer06a.pdf> K. Crammer, O. Dekel, J. Keshat, S. Shalev-Shwartz, Y. Singer - JMLR (2006)

Parameters:

  • alpha (float) – Maximum step size (regularization)

  • eta0 (float) – The initial learning rate for the ‘adaptive’ schedules.

  • loss (str) – The loss function to be used: hinge: equivalent to PA-I in the reference paper. squared_hinge: equivalent to PA-II in the reference paper.

  • label_weights (dict[str, float] | None) – An optional dictionary of label names and their relative weight. e.g., {‘Go’: 31.0, ‘Stop’: 0.5} If this is None then settings_path must be provided and the pre-trained model

  • settings_path (str | None) – Path to the stored sklearn model pkl file.

Return type:

Generator[AxisArray | SampleMessage, ClassifierMessage | None, None]

Returns:

Generator that accepts SampleMessage for incremental training (partial_fit) and yields None, or AxisArray for inference (predict) and yields a ClassifierMessage.

Classes

class SGDDecoder(*args, settings=None, **kwargs)[source]#

Bases: GenAxisArray

Parameters:

settings (Settings | None)

SETTINGS#

alias of SGDDecoderSettings

INPUT_SAMPLE = InputStream:unlocated[<class 'ezmsg.baseproc.util.message.SampleMessage'>]()#
construct_generator()[source]#
async on_sample(msg)[source]#
Return type:

None

Parameters:

msg (SampleMessage)

class SGDDecoderSettings(alpha=1e-05, eta0=0.0003, loss='hinge', label_weights=None, settings_path=None)[source]#

Bases: Settings

Parameters:
alpha: float = 1e-05#
eta0: float = 0.0003#
loss: str = 'hinge'#
label_weights: dict[str, float] | None = None#
settings_path: str | None = None#
__init__(alpha=1e-05, eta0=0.0003, loss='hinge', label_weights=None, settings_path=None)#
Parameters:
Return type:

None

sgd_decoder(alpha=1.5e-05, eta0=1e-07, loss='squared_hinge', label_weights=None, settings_path=None)[source]#

Passive Aggressive Classifier Online Passive-Aggressive Algorithms <http://jmlr.csail.mit.edu/papers/volume7/crammer06a/crammer06a.pdf> K. Crammer, O. Dekel, J. Keshat, S. Shalev-Shwartz, Y. Singer - JMLR (2006)

Parameters:

  • alpha (float) – Maximum step size (regularization)

  • eta0 (float) – The initial learning rate for the ‘adaptive’ schedules.

  • loss (str) – The loss function to be used: hinge: equivalent to PA-I in the reference paper. squared_hinge: equivalent to PA-II in the reference paper.

  • label_weights (dict[str, float] | None) – An optional dictionary of label names and their relative weight. e.g., {‘Go’: 31.0, ‘Stop’: 0.5} If this is None then settings_path must be provided and the pre-trained model

  • settings_path (str | None) – Path to the stored sklearn model pkl file.

Return type:

Generator[AxisArray | SampleMessage, ClassifierMessage | None, None]

Returns:

Generator that accepts SampleMessage for incremental training (partial_fit) and yields None, or AxisArray for inference (predict) and yields a ClassifierMessage.

class SGDDecoderSettings(alpha=1e-05, eta0=0.0003, loss='hinge', label_weights=None, settings_path=None)[source]#

Bases: Settings

Parameters:
alpha: float = 1e-05#
eta0: float = 0.0003#
loss: str = 'hinge'#
label_weights: dict[str, float] | None = None#
settings_path: str | None = None#
__init__(alpha=1e-05, eta0=0.0003, loss='hinge', label_weights=None, settings_path=None)#
Parameters:
Return type:

None

class SGDDecoder(*args, settings=None, **kwargs)[source]#

Bases: GenAxisArray

Parameters:

settings (Settings | None)

SETTINGS#

alias of SGDDecoderSettings

INPUT_SAMPLE = InputStream:unlocated[<class 'ezmsg.baseproc.util.message.SampleMessage'>]()#
construct_generator()[source]#
async on_sample(msg)[source]#
Return type:

None

Parameters:

msg (SampleMessage)