ezmsg.learn.dim_reduce.adaptive_decomp

Classes

class AdaptiveDecompSettings(axis='!time', n_components=2)

Bases: Settings

Parameters:

• axis (str)

• n_components (int)

axis: str = '!time'

n_components: int = 2

__init__(axis='!time', n_components=2)

Parameters:

• axis (str)

• n_components (int)

Return type:

None

class AdaptiveDecompState

Bases: object

template: AxisArray | None = None

axis_groups: tuple[str, list[str], list[str]] | None = None

estimator: Any = None

class AdaptiveDecompTransformer(*args, **kwargs)

Bases: BaseAdaptiveTransformer[AdaptiveDecompSettings, AxisArray, AxisArray, AdaptiveDecompState], Generic[EstimatorType]

Base class for adaptive decomposition transformers. See IncrementalPCATransformer and MiniBatchNMFTransformer for concrete implementations.

Note that for these classes, adaptation is not automatic. The user must call partial_fit on the transformer. For automated adaptation, see IncrementalDecompTransformer.

__init__(*args, **kwargs)

classmethod get_message_type(dir)

Return type:

Type[AxisArray]

Parameters:

dir (str)

classmethod get_estimator_type()

Return type:

Type[TypeVar(EstimatorType, bound= Union[IncrementalPCA, MiniBatchNMF])]

partial_fit(message)

Return type:

None

Parameters:

message (AxisArray)

class BaseAdaptiveDecompUnit(*args, settings=None, **kwargs)

Bases: BaseAdaptiveTransformerUnit[SettingsType, AxisArray, AxisArray, TransformerType], Generic[SettingsType, TransformerType]

Parameters:

settings (Settings | None)

INPUT_SAMPLE = InputStream:unlocated[<class 'ezmsg.util.messages.axisarray.AxisArray'>]()

async on_sample(msg)

Return type:

None

Parameters:

msg (AxisArray)

class IncrementalPCASettings(axis='!time', n_components=2, whiten=False, batch_size=None)

Bases: AdaptiveDecompSettings

Parameters:

• axis (str)

• n_components (int)

• whiten (bool)

• batch_size (int | None)

whiten: bool = False

batch_size: Optional[int] = None

__init__(axis='!time', n_components=2, whiten=False, batch_size=None)

Parameters:

• axis (str)

• n_components (int)

• whiten (bool)

• batch_size (int | None)

Return type:

None

class IncrementalPCATransformer(*args, **kwargs)

Bases: AdaptiveDecompTransformer[IncrementalPCA]

class IncrementalPCAUnit(*args, settings=None, **kwargs)

Bases: BaseAdaptiveDecompUnit[IncrementalPCASettings, IncrementalPCATransformer]

Parameters:

settings (Settings | None)

SETTINGS

alias of IncrementalPCASettings

class MiniBatchNMFSettings(axis='!time', n_components=2, init='random', batch_size=1024, beta_loss='frobenius', tol=0.0001, max_no_improvement=None, max_iter=200, alpha_W=0.0, alpha_H='same', l1_ratio=0.0, forget_factor=0.7)

Bases: AdaptiveDecompSettings

Parameters:

• axis (str)

• n_components (int)

• init (str | None)

• batch_size (int)

• beta_loss (str | float)

• tol (float)

• max_no_improvement (int | None)

• max_iter (int)

• alpha_W (float)

• alpha_H (float | str)

• l1_ratio (float)

• forget_factor (float)

init: Optional[str] = 'random'

‘random’, ‘nndsvd’, ‘nndsvda’, ‘nndsvdar’, ‘custom’, or None

batch_size: int = 1024

batch_size is used only when doing a full fit (i.e., a reset), or as the exponent to forget_factor, where a very small batch_size will cause the model to update more slowly. It is better to set batch_size to a larger number than the expected chunk size and instead use forget_factor to control the learning rate.

beta_loss: Union[str, float] = 'frobenius'

‘frobenius’, ‘kullback-leibler’, ‘itakura-saito’ Note that values different from ‘frobenius’

(or 2) and ‘kullback-leibler’ (or 1) lead to significantly slower fits. Note that for beta_loss <= 0 (or ‘itakura-saito’), the input matrix X cannot contain zeros.

tol: float = 0.0001

max_no_improvement: Optional[int] = None

max_iter: int = 200

alpha_W: float = 0.0

alpha_H: Union[float, str] = 'same'

l1_ratio: float = 0.0

forget_factor: float = 0.7

__init__(axis='!time', n_components=2, init='random', batch_size=1024, beta_loss='frobenius', tol=0.0001, max_no_improvement=None, max_iter=200, alpha_W=0.0, alpha_H='same', l1_ratio=0.0, forget_factor=0.7)

Parameters:

• axis (str)

• n_components (int)

• init (str | None)

• batch_size (int)

• beta_loss (str | float)

• tol (float)

• max_no_improvement (int | None)

• max_iter (int)

• alpha_W (float)

• alpha_H (float | str)

• l1_ratio (float)

• forget_factor (float)

Return type:

None

class MiniBatchNMFTransformer(*args, **kwargs)

Bases: AdaptiveDecompTransformer[MiniBatchNMF]

class MiniBatchNMFUnit(*args, settings=None, **kwargs)

Bases: BaseAdaptiveDecompUnit[MiniBatchNMFSettings, MiniBatchNMFTransformer]

Parameters:

settings (Settings | None)

SETTINGS

alias of MiniBatchNMFSettings

class AdaptiveDecompSettings(axis='!time', n_components=2)

Bases: Settings

Parameters:

• axis (str)

• n_components (int)

axis: str = '!time'

n_components: int = 2

__init__(axis='!time', n_components=2)

Parameters:

• axis (str)

• n_components (int)

Return type:

None

class AdaptiveDecompState

Bases: object

template: AxisArray | None = None

axis_groups: tuple[str, list[str], list[str]] | None = None

estimator: Any = None

class AdaptiveDecompTransformer(*args, **kwargs)

Bases: BaseAdaptiveTransformer[AdaptiveDecompSettings, AxisArray, AxisArray, AdaptiveDecompState], Generic[EstimatorType]

Base class for adaptive decomposition transformers. See IncrementalPCATransformer and MiniBatchNMFTransformer for concrete implementations.

Note that for these classes, adaptation is not automatic. The user must call partial_fit on the transformer. For automated adaptation, see IncrementalDecompTransformer.

__init__(*args, **kwargs)

classmethod get_message_type(dir)

Return type:

Type[AxisArray]

Parameters:

dir (str)

classmethod get_estimator_type()

Return type:

Type[TypeVar(EstimatorType, bound= Union[IncrementalPCA, MiniBatchNMF])]

partial_fit(message)

Return type:

None

Parameters:

message (AxisArray)

class IncrementalPCASettings(axis='!time', n_components=2, whiten=False, batch_size=None)

Bases: AdaptiveDecompSettings

Parameters:

• axis (str)

• n_components (int)

• whiten (bool)

• batch_size (int | None)

whiten: bool = False

batch_size: Optional[int] = None

__init__(axis='!time', n_components=2, whiten=False, batch_size=None)

Parameters:

• axis (str)

• n_components (int)

• whiten (bool)

• batch_size (int | None)

Return type:

None

class IncrementalPCATransformer(*args, **kwargs)

Bases: AdaptiveDecompTransformer[IncrementalPCA]

class MiniBatchNMFSettings(axis='!time', n_components=2, init='random', batch_size=1024, beta_loss='frobenius', tol=0.0001, max_no_improvement=None, max_iter=200, alpha_W=0.0, alpha_H='same', l1_ratio=0.0, forget_factor=0.7)

Bases: AdaptiveDecompSettings

Parameters:

• axis (str)

• n_components (int)

• init (str | None)

• batch_size (int)

• beta_loss (str | float)

• tol (float)

• max_no_improvement (int | None)

• max_iter (int)

• alpha_W (float)

• alpha_H (float | str)

• l1_ratio (float)

• forget_factor (float)

init: Optional[str] = 'random'

‘random’, ‘nndsvd’, ‘nndsvda’, ‘nndsvdar’, ‘custom’, or None

batch_size: int = 1024

batch_size is used only when doing a full fit (i.e., a reset), or as the exponent to forget_factor, where a very small batch_size will cause the model to update more slowly. It is better to set batch_size to a larger number than the expected chunk size and instead use forget_factor to control the learning rate.

beta_loss: Union[str, float] = 'frobenius'

‘frobenius’, ‘kullback-leibler’, ‘itakura-saito’ Note that values different from ‘frobenius’

(or 2) and ‘kullback-leibler’ (or 1) lead to significantly slower fits. Note that for beta_loss <= 0 (or ‘itakura-saito’), the input matrix X cannot contain zeros.

tol: float = 0.0001

max_no_improvement: Optional[int] = None

max_iter: int = 200

alpha_W: float = 0.0

alpha_H: Union[float, str] = 'same'

l1_ratio: float = 0.0

forget_factor: float = 0.7

__init__(axis='!time', n_components=2, init='random', batch_size=1024, beta_loss='frobenius', tol=0.0001, max_no_improvement=None, max_iter=200, alpha_W=0.0, alpha_H='same', l1_ratio=0.0, forget_factor=0.7)

Parameters:

• axis (str)

• n_components (int)

• init (str | None)

• batch_size (int)

• beta_loss (str | float)

• tol (float)

• max_no_improvement (int | None)

• max_iter (int)

• alpha_W (float)

• alpha_H (float | str)

• l1_ratio (float)

• forget_factor (float)

Return type:

None

class MiniBatchNMFTransformer(*args, **kwargs)

Bases: AdaptiveDecompTransformer[MiniBatchNMF]

class BaseAdaptiveDecompUnit(*args, settings=None, **kwargs)

Bases: BaseAdaptiveTransformerUnit[SettingsType, AxisArray, AxisArray, TransformerType], Generic[SettingsType, TransformerType]

Parameters:

settings (Settings | None)

INPUT_SAMPLE = InputStream:unlocated[<class 'ezmsg.util.messages.axisarray.AxisArray'>]()

async on_sample(msg)

Return type:

None

Parameters:

msg (AxisArray)

class IncrementalPCAUnit(*args, settings=None, **kwargs)

Bases: BaseAdaptiveDecompUnit[IncrementalPCASettings, IncrementalPCATransformer]

Parameters:

settings (Settings | None)

SETTINGS

alias of IncrementalPCASettings

class MiniBatchNMFUnit(*args, settings=None, **kwargs)

Bases: BaseAdaptiveDecompUnit[MiniBatchNMFSettings, MiniBatchNMFTransformer]

Parameters:

settings (Settings | None)

SETTINGS

alias of MiniBatchNMFSettings