NormalizedStudentAffinity

class torchdr.NormalizedStudentAffinity(degrees_of_freedom: float = 1.0, metric: str = 'sqeuclidean', zero_diag: bool = True, device: str = 'auto', keops: bool = False, verbose: bool = False, normalization_dim: int | Tuple[int] = (0, 1))[source]

Bases: LogAffinity

Compute the Student affinity matrix which can be normalized along a dimension.

Its expression is given by:

\[\left(1 + \frac{\mathbf{C}}{\nu}\right)^{-\frac{\nu + 1}{2}}\]

where \(\nu > 0\) is the degrees of freedom parameter. The affinity can be normalized according to the specified normalization dimension.

Parameters:

  • degrees_of_freedom (int, optional) – Degrees of freedom for the Student-t distribution.

  • metric (str, optional) – Metric to use for pairwise distances computation.

  • zero_diag (bool, optional) – Whether to set the diagonal of the affinity matrix to zero.

  • device (str, optional) – Device to use for computations.

  • keops (bool, optional) – Whether to use KeOps for computations.

  • verbose (bool, optional) – Verbosity.

  • normalization_dim (int or Tuple[int], optional) – Dimension along which to normalize the affinity matrix. Default is (0, 1)