Bayes Error Rate

Tutorials

Check out this tutorial to begin using the BER class

Bayes Error Rate Tutorial

How To Guides

There are currently no how to’s for BER. If there are scenarios that you want us to explain, contact us!

DataEval API

class dataeval.metrics.BER(data: ArrayLike, labels: ArrayLike, method: Literal['MST', 'KNN'] = 'KNN', k: int = 1)

An estimator for Multi-class Bayes Error Rate using FR or KNN test statistic basis

Parameters:

• data (np.ndarray) – Array of images or image embeddings

• labels (np.ndarray) – Array of labels for each image or image embedding

• method (Literal["MST", "KNN"], default "KNN") – Method to use when estimating the Bayes error rate

• k (int, default 1) – number of nearest neighbors for KNN estimator – ignored by MST estimator

See also

Learning to Bound the Multi-class Bayes Error (Th. 3 and Th. 4)

evaluate() → Dict[str, float]

Calculates the Bayes Error Rate estimate using the provided method

Returns:

berfloat

The estimated lower bounds of the Bayes Error Rate

ber_lowerfloat

The estimated upper bounds of the Bayes Error Rate

Return type:

Dict[str, float]

Raises:

ValueError – If unique classes M < 2