coco_pipe.transforms.subject_alignment
======================================

.. py:module:: coco_pipe.transforms.subject_alignment

.. autoapi-nested-parse::

   Experimental: subject-axis removal for frozen embedding vectors.



Attributes
----------

.. autoapisummary::

   coco_pipe.transforms.subject_alignment.VECTOR_TRANSFORMS
   coco_pipe.transforms.subject_alignment.TOKEN_TRANSFORMS


Classes
-------

.. autoapisummary::

   coco_pipe.transforms.subject_alignment.SubjectAlignment
   coco_pipe.transforms.subject_alignment.LeaceEraser
   coco_pipe.transforms.subject_alignment.EuclideanAlign
   coco_pipe.transforms.subject_alignment.RiemannAlign


Functions
---------

.. autoapisummary::

   coco_pipe.transforms.subject_alignment.tokens_to_covariances
   coco_pipe.transforms.subject_alignment.make_subject_transform


Module Contents
---------------

.. py:class:: SubjectAlignment

   Bases: :py:obj:`sklearn.base.BaseEstimator`, :py:obj:`sklearn.base.TransformerMixin`


   Base class for subject-alignment estimators.


   .. py:attribute:: name
      :value: 'identity'



   .. py:attribute:: fold_local
      :type:  bool
      :value: False



   .. py:method:: fingerprint()


.. py:class:: LeaceEraser(*, shrinkage = True, degenerate_frac = 0.95)

   Bases: :py:obj:`SubjectAlignment`


   Fold-local LEACE projection of the between-subject linear subspace.


   .. py:attribute:: name
      :value: 'leace'



   .. py:attribute:: fold_local
      :value: True



   .. py:attribute:: shrinkage
      :value: True



   .. py:attribute:: degenerate_frac
      :value: 0.95



   .. py:method:: fit(X, y=None, groups=None)


   .. py:method:: transform(X, groups=None)


   .. py:method:: fit_transform(X, y=None, groups=None, **fit_params)

      Fit the LEACE projector and transform the same observations.



.. py:class:: EuclideanAlign(mode = 'coral', *, shrinkage = True)

   Bases: :py:obj:`SubjectAlignment`


   Per-subject mean or CORAL alignment, including unseen subjects.


   .. py:attribute:: mode
      :value: 'coral'



   .. py:attribute:: name
      :value: 'ea_coral'



   .. py:attribute:: shrinkage
      :value: True



   .. py:method:: fit(X, y=None, groups=None)


   .. py:method:: transform(X, groups=None)


   .. py:method:: fit_transform(X, y=None, groups=None, **fit_params)

      Fit subject statistics and transform the same observations.



.. py:function:: tokens_to_covariances(tokens, *, shrinkage = 0.1)

   Convert ``(window, token, feature)`` tensors to regularized SPD matrices.


.. py:class:: RiemannAlign(*, shrinkage = 0.1)

   Bases: :py:obj:`SubjectAlignment`


   Per-subject Riemannian recentering followed by tangent projection.


   .. py:attribute:: name
      :value: 'ra'



   .. py:attribute:: shrinkage
      :value: 0.1



   .. py:method:: fit(X, y=None, groups=None)


   .. py:method:: transform(X, groups=None)


   .. py:method:: fit_transform(X, y=None, groups=None, **fit_params)

      Fit subject references and transform the same token windows.



.. py:data:: VECTOR_TRANSFORMS

.. py:data:: TOKEN_TRANSFORMS

.. py:function:: make_subject_transform(name, **params)

   Instantiate a registered subject transform by name.