TemplateClassif

class scio.scores.classification.TemplateClassif(*, act_norm=None, mode='raw', param_with_a_default_value='default value', param_with_no_default_value=?)

Bases: BaseScoreClassif

Template for classification.

The source code of this class is a good starting point if you wish to implement a new score.

Parameters:

• param_with_a_default_value (str) – A parameter with a default value. Defaults to "default value".

• param_with_no_default_value (int) – A parameter with no default value.

• mode – See BaseScoreClassif.

• act_norm – See BaseScore.

References

[AA00]

Template Author and Template Author. Template reference. Template journal, 1900.

Useful methods defined here

calibrate(calib_data, calib_labels)	Calibrate the scoring algorithm with In-Distribution data.
get_conformity(inputs)	Compute output and associated conformity at inference.

calibrate(calib_data, calib_labels)

Calibrate the scoring algorithm with In-Distribution data.

Example

def calibrate(self, calib_data: Tensor, calib_labels: Tensor) -> None:
    """Calibrate the scoring algorithm with In-Distribution data."""
    out = self.rnet(calib_data)  # records activations
    activations = self.activations()
    self.calibration_statistics = ...

get_conformity(inputs)

Compute output and associated conformity at inference.

Example

def get_conformity(self, inputs: Tensor) -> tuple[Tensor, Tensor]:
    """Compute output and associated conformity at inference."""
    out = self.rnet(inputs)  # records activations
    conformity = ...
    return out, conformity