Move QuantumClassifierWithDefaultRiemanianPipeline to pipelines module (

#159)

* Create pipelines.py

* add pipeline module and move QuantumClassifierWithRiemannianPipeline to it

* [pre-commit.ci] auto fixes from pre-commit.com hooks

* flake8

* [pre-commit.ci] auto fixes from pre-commit.com hooks

* flake8

* fix dockerfile

* fix dockerfile

* fix import

* [pre-commit.ci] auto fixes from pre-commit.com hooks

* flake8

* [pre-commit.ci] auto fixes from pre-commit.com hooks

* typo

* fix import in example

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Dockerfile

@@ -22,10 +22,10 @@ RUN mkdir /root/mne_data
 RUN mkdir /home/mne_data
 
 ## Workaround for firestore
-RUN pip install protobuf==4.23.2
+RUN pip install protobuf==4.23.3
 RUN pip install google_cloud_firestore==2.11.1
 ### Missing __init__ file in protobuf
-RUN touch /usr/local/lib/python3.8/site-packages/protobuf-4.23.2-py3.8.egg/google/__init__.py
+RUN touch /usr/local/lib/python3.8/site-packages/protobuf-4.23.3-py3.8-linux-x86_64.egg/google/__init__.py
 ## google.cloud.location is never used in these files, and is missing in path.
 RUN sed -i 's/from google.cloud.location import locations_pb2//g' '/usr/local/lib/python3.8/site-packages/google_cloud_firestore-2.11.1-py3.8.egg/google/cloud/firestore_v1/services/firestore/client.py'
 RUN sed -i 's/from google.cloud.location import locations_pb2//g' '/usr/local/lib/python3.8/site-packages/google_cloud_firestore-2.11.1-py3.8.egg/google/cloud/firestore_v1/services/firestore/transports/base.py'

doc/api.rst

@@ -17,6 +17,17 @@ Classification
     QuanticSVM
     QuanticVQC
     QuanticMDM
+
+
+Pipelines
+---------
+.. _pipelines_api:
+.. currentmodule:: pyriemann_qiskit.pipelines
+
+.. autosummary::
+    :toctree: generated/
+    :template: class.rst
+
     QuantumClassifierWithDefaultRiemannianPipeline
 
 

examples/ERP/firebase_moabb.py

@@ -40,7 +40,7 @@
 from moabb.datasets import bi2012
 from moabb.evaluations import WithinSessionEvaluation
 from moabb.paradigms import P300
-from pyriemann_qiskit.classification import (
+from pyriemann_qiskit.pipelines import (
     QuantumClassifierWithDefaultRiemannianPipeline,
 )
 from sklearn.decomposition import PCA

examples/ERP/plot_classify_P300_bi.py

@@ -33,7 +33,7 @@
 from moabb.datasets import bi2012
 from moabb.evaluations import WithinSessionEvaluation
 from moabb.paradigms import P300
-from pyriemann_qiskit.classification import (
+from pyriemann_qiskit.pipelines import (
     QuantumClassifierWithDefaultRiemannianPipeline,
 )
 from sklearn.decomposition import PCA

examples/ERP/plot_compare_dim_red.py

@@ -20,7 +20,7 @@
 # License: BSD (3-clause)
 
 from pyriemann_qiskit.datasets import get_mne_sample
-from pyriemann_qiskit.classification import (
+from pyriemann_qiskit.pipelines import (
     QuantumClassifierWithDefaultRiemannianPipeline,
 )
 from pyriemann_qiskit.utils.filtering import NaiveDimRed

pyriemann_qiskit/__init__.py

@@ -1,4 +1,5 @@
 from ._version import __version__
 from . import classification
+from . import pipelines
 
-__all__ = ["__version__", "classification"]
+__all__ = ["__version__", "classification", "pipelines"]

pyriemann_qiskit/classification.py

@@ -1,8 +1,6 @@
 """Module for classification function."""
 import numpy as np
-from sklearn.base import BaseEstimator, ClassifierMixin, TransformerMixin
-from sklearn.decomposition import PCA
-from sklearn.pipeline import make_pipeline
+from sklearn.base import BaseEstimator, ClassifierMixin
 from sklearn.svm import SVC
 from qiskit_ibm_provider import IBMProvider, least_busy
 from qiskit.utils import QuantumInstance, algorithm_globals
@@ -13,9 +11,7 @@
 import logging
 from .utils.hyper_params_factory import gen_zz_feature_map, gen_two_local, get_spsa
 from .utils import get_provider, get_devices, get_simulator
-from pyriemann.estimation import XdawnCovariances
 from pyriemann.classification import MDM
-from pyriemann.tangentspace import TangentSpace
 from pyriemann_qiskit.datasets import get_feature_dimension
 from pyriemann_qiskit.utils import (
     ClassicalOptimizer,
@@ -591,239 +587,3 @@ def predict(self, X):
         """
         labels = self._predict(X)
         return self._map_0_1_to_classes(labels)
-
-
-class QuantumClassifierWithDefaultRiemannianPipeline(
-    BaseEstimator, ClassifierMixin, TransformerMixin
-):
-
-    """Default pipeline with Riemann Geometry and a quantum classifier.
-
-    Projects the data into the tangent space of the Riemannian manifold
-    and applies quantum classification.
-
-    The type of quantum classification (quantum SVM or VQC) depends on
-    the value of the parameters.
-
-    Data are entangled using a ZZFeatureMap. A SPSA optimizer and a two-local
-    circuits are used in addition when the VQC is selected.
-
-
-
-    Parameters
-    ----------
-    nfilter : int (default: 1)
-        The number of filter for the xDawnFilter.
-        The number of components selected is 2 x nfilter.
-    dim_red : TransformerMixin (default: PCA())
-        A transformer that will reduce the dimension of the feature,
-        after the data are projected into the tangent space.
-    gamma : float | None (default:None)
-        Used as input for sklearn rbf_kernel which is used internally.
-        See [1]_ for more information about gamma.
-    C : float (default: 1.0)
-        Regularization parameter. The strength of the regularization is
-        inversely proportional to C. Must be strictly positive.
-        Note, if pegasos is enabled you may want to consider
-        larger values of C.
-    max_iter: int | None (default: None)
-        number of steps in Pegasos or SVC.
-        If None, respective default values for Pegasos and (Q)SVC
-        are used. The default value for Pegasos is 1000.
-        For (Q)SVC it is -1 (that is not limit).
-    shots : int | None (default: 1024)
-        Number of repetitions of each circuit, for sampling.
-        If None, classical computation will be performed.
-    feature_entanglement : str | list[list[list[int]]] | \
-                   Callable[int, list[list[list[int]]]]
-        Specifies the entanglement structure for the ZZFeatureMap.
-        Entanglement structure can be provided with indices or string.
-        Possible string values are: 'full', 'linear', 'circular' and 'sca'.
-        Consult [2]_ for more details on entanglement structure.
-    feature_reps : int (default: 2)
-        The number of repeated circuits for the ZZFeatureMap,
-        greater or equal to 1.
-    spsa_trials : int (default: 40)
-        Maximum number of iterations to perform using SPSA optimizer.
-    two_local_reps : int (default: 3)
-        The number of repetition for the two-local cricuit.
-    params: Dict (default: {})
-        Additional parameters to pass to the nested instance
-        of the quantum classifier.
-        See QuanticClassifierBase, QuanticVQC and QuanticSVM for
-        a complete list of the parameters.
-
-    Notes
-    -----
-    .. versionadded:: 0.0.1
-
-    See Also
-    --------
-    XdawnCovariances
-    TangentSpace
-    gen_zz_feature_map
-    gen_two_local
-    get_spsa
-    QuanticVQC
-    QuanticSVM
-    QuanticClassifierBase
-
-    References
-    ----------
-    .. [1] Available from: \
-        https://scikit-learn.org/stable/modules/generated/sklearn.metrics.pairwise.rbf_kernel.html
-
-    .. [2] \
-        https://qiskit.org/documentation/stable/0.36/stubs/qiskit.circuit.library.NLocal.html
-
-    """
-
-    def __init__(
-        self,
-        nfilter=1,
-        dim_red=PCA(),
-        gamma="scale",
-        C=1.0,
-        max_iter=None,
-        shots=1024,
-        feature_entanglement="full",
-        feature_reps=2,
-        spsa_trials=None,
-        two_local_reps=None,
-        params={},
-    ):
-        self.nfilter = nfilter
-        self.dim_red = dim_red
-        self.gamma = gamma
-        self.C = C
-        self.max_iter = max_iter
-        self.shots = shots
-        self.feature_entanglement = feature_entanglement
-        self.feature_reps = feature_reps
-        self.spsa_trials = spsa_trials
-        self.two_local_reps = two_local_reps
-        self.params = params
-
-        is_vqc = spsa_trials and two_local_reps
-        is_quantum = shots is not None
-
-        feature_map = gen_zz_feature_map(feature_reps, feature_entanglement)
-        # verbose is passed as an additional parameter to quantum classifiers.
-        self.verbose = "verbose" in params and params["verbose"]
-        if is_vqc:
-            self._log("QuanticVQC chosen.")
-            clf = QuanticVQC(
-                optimizer=get_spsa(spsa_trials),
-                gen_var_form=gen_two_local(two_local_reps),
-                gen_feature_map=feature_map,
-                shots=self.shots,
-                quantum=is_quantum,
-                **params
-            )
-        else:
-            self._log("QuanticSVM chosen.")
-            clf = QuanticSVM(
-                quantum=is_quantum,
-                gamma=gamma,
-                C=C,
-                max_iter=max_iter,
-                gen_feature_map=feature_map,
-                shots=shots,
-                **params
-            )
-
-        self._pipe = make_pipeline(
-            XdawnCovariances(nfilter=nfilter), TangentSpace(), dim_red, clf
-        )
-
-    def _log(self, trace):
-        if self.verbose:
-            print("[QuantumClassifierWithDefaultRiemannianPipeline] ", trace)
-
-    def fit(self, X, y):
-        """Train the riemann quantum classifier.
-
-        Parameters
-        ----------
-        X : ndarray, shape (n_trials, n_channels, n_times)
-            ndarray of trials.
-        y : ndarray, shape (n_samples,)
-            Target vector relative to X.
-
-        Returns
-        -------
-        self : QuantumClassifierWithDefaultRiemannianPipeline instance
-            The QuantumClassifierWithDefaultRiemannianPipeline instance
-        """
-
-        self.classes_ = np.unique(y)
-        self._pipe.fit(X, y)
-        return self
-
-    def score(self, X, y):
-        """Return the accuracy.
-        You might want to use a different metric by using sklearn
-        cross_val_score
-
-        Parameters
-        ----------
-        X : ndarray, shape (n_trials, n_channels, n_times)
-            ndarray of trials.
-        y : ndarray, shape (n_trials,)
-            Predicted target vector relative to X.
-
-        Returns
-        -------
-        accuracy : double
-            Accuracy of predictions from X with respect y.
-        """
-        return self._pipe.score(X, y)
-
-    def predict(self, X):
-        """get the predictions.
-
-        Parameters
-        ----------
-        X : ndarray, shape (n_trials, n_channels, n_times)
-            ndarray of trials.
-
-        Returns
-        -------
-        pred : ndarray of int, shape (n_trials, 1)
-            Class labels for samples in X.
-        """
-        return self._pipe.predict(X)
-
-    def predict_proba(self, X):
-        """Return the probabilities associated with predictions.
-
-        Parameters
-        ----------
-        X : ndarray, shape (n_trials, n_channels, n_times)
-            ndarray of trials.
-
-        Returns
-        -------
-        prob : ndarray, shape (n_samples, n_classes)
-            prob[n, 0] == True if the nth sample is assigned to 1st class;
-            prob[n, 1] == True if the nth sample is assigned to 2nd class.
-        """
-
-        return self._pipe.predict_proba(X)
-
-    def transform(self, X):
-        """Transform the data into feature vectors.
-
-        Parameters
-        ----------
-        X : ndarray, shape (n_trials, n_channels, n_times)
-            ndarray of trials.
-
-        Returns
-        -------
-        dist : ndarray, shape (n_trials, n_ts)
-            the tangent space projection of the data.
-            the dimension of the feature vector depends on
-            `n_filter` and `dim_red`.
-        """
-        return self._pipe.transform(X)