biolord.BiolordClassifyModule#
- class biolord.BiolordClassifyModule(categorical_attributes_missing=None, classify_all=False, logits=False, bias=True, classification_penalty=0.1, classifier_penalty=0.0001, classifier_nn_width=128, classifier_nn_depth=2, classifier_dropout_rate=0.1, loss_regression='normal', **kwargs)[source]#
The
biolord-classify
module.A
BiolordModule
accompanied by regressors for ordered classes and classifiers for categorical classes.- Parameters:
categorical_attributes_missing (
Optional
[dict
[str
,str
]]) – Categorical categories representing un-labeled cells.classify_all (
bool
) – Whether to classify all classes or only semi-supervised classes.logits (
bool
) – Classifier output type.bias (
bool
) – Whether to add bias to the regressor.classification_penalty (
float
) – Classification penalty strength.classifier_nn_width (
int
) – Classifier’s layer width.classifier_nn_depth (
int
) – Classifier’s number of layers.classifier_dropout_rate (
float
) – Classifier’s dropout rate.loss_regression (
Literal
['normal'
,'mse'
]) – Loss function for regressorskwargs (
Any
) – Keyword arguments forBiolordModule
.
Attributes table#
Methods table#
|
Run classification. |
|
Compute the loss. |