Chores/plot distance visualization (#284)

* plot distances_visualization on Ci

* fix hardcoded parameter in plot_financial_data

* fix label in plot_classifier_comparison

* apply noplot convention

* remove tight_layout (not necessary)

* add alias module

* add alias file

* missing line at end of file

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

* fix flake8

* fix warning

* remove alias module

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

* missing in previous commit

---------

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

examples/MI/helpers/alias.py

@@ -0,0 +1,5 @@
+from pyriemann.estimation import ERPCovariances
+from pyriemann.classification import MDM
+from sklearn.pipeline import make_pipeline
+
+ERPCov_MDM = make_pipeline(ERPCovariances(estimator="lwf"), MDM())

examples/MI/classify_alexmi_with_quantum_pipeline.py → examples/MI/noplot_classify_alexmi_with_quantum_pipeline.py

@@ -21,15 +21,12 @@
 from moabb.paradigms import MotorImagery
 
 # inject cpm distance and mean to pyriemann (if not done already)
+from helpers.alias import ERPCov_MDM
 from pyriemann_qiskit.utils import distance, mean  # noqa
 from pyriemann_qiskit.pipelines import (
     QuantumMDMWithRiemannianPipeline,
 )
 
-from sklearn.pipeline import make_pipeline
-from pyriemann.estimation import ERPCovariances
-from pyriemann.classification import MDM
-
 print(__doc__)
 
 ##############################################################################
@@ -73,7 +70,7 @@
 )
 
 # Classical baseline for evaluation
-pipelines["R-MDM"] = make_pipeline(ERPCovariances(estimator="lwf"), MDM())
+pipelines["R-MDM"] = ERPCov_MDM
 
 ##############################################################################
 # Run evaluation

examples/MI/multiclass_classification.py → examples/MI/noplot_multiclass_classification.py

@@ -13,14 +13,12 @@
 # Modified from plot_classify_EEG_quantum_svm
 # License: BSD (3-clause)
 
+from helpers.alias import ERPCov_MDM
 from pyriemann_qiskit.datasets import get_mne_sample
 from pyriemann_qiskit.pipelines import (
     QuantumClassifierWithDefaultRiemannianPipeline,
     QuantumMDMWithRiemannianPipeline,
 )
-from pyriemann.estimation import ERPCovariances
-from pyriemann.classification import MDM
-from sklearn.pipeline import make_pipeline
 from sklearn.model_selection import train_test_split
 from sklearn.metrics import (
     confusion_matrix,
@@ -73,7 +71,7 @@
 quantum_mdm = QuantumMDMWithRiemannianPipeline()
 
 # Pipeline 5
-mdm = make_pipeline(ERPCovariances(estimator="lwf"), MDM())
+mdm = ERPCov_MDM
 
 classifiers = [vqc, quantum_svm, classical_svm, quantum_mdm, mdm]
 

examples/other_datasets/plot_financial_data.py

@@ -251,7 +251,7 @@ def fit(self, X, y):
     def transform(self, X):
         if not self.process:
             return X
-        return Whitening(dim_red={"n_components": 4}).fit_transform(X)
+        return Whitening(dim_red={"n_components": self.n_components}).fit_transform(X)
 
 
 # Create a RandomForest for baseline comparison of direct classification:

examples/toys_dataset/plot_classifier_comparison.py

@@ -23,7 +23,11 @@
     generate_linearly_separable_dataset,
     generate_qiskit_dataset,
 )
-from pyriemann_qiskit.classification import QuanticSVM
+from pyriemann_qiskit.classification import (
+    QuanticSVM,
+    # uncomment to run comparison with QuanticVQC (disabled for CI/CD)
+    # QuanticVQC
+)
 
 # uncomment to run comparison with QuanticVQC (disabled for CI/CD)
 # from pyriemann_qiskit.classification import QuanticVQC
@@ -36,14 +40,20 @@
 
 h = 0.02  # step size in the mesh
 labels = (0, 1)
-names = ["Linear SVM", "RBF SVM", "VQC", "QSVM"]
+names = [
+    "Linear SVM",
+    "RBF SVM",
+    # uncomment to run comparison with QuanticVQC (disabled for CI/CD)
+    # "VQC",
+    "QSVM",
+]
 
 classifiers = [
     SVC(kernel="linear", C=0.025),
     SVC(gamma="auto", C=0.001),
     # uncomment to run comparison with QuanticVQC (disabled for CI/CD)
     # QuanticVQC(),
-    QuanticSVM(quantum=False),
+    QuanticSVM(quantum=False),  # quantum=False for CI
 ]
 
 # Warning: There is a known convergence issue with QSVM

examples/toys_dataset/plot_learning_curve.py

@@ -81,5 +81,4 @@
 axe.set_ylabel("Evaluated values (MDM)")
 axe.set_xlabel("Evaluations")
 
-plt.tight_layout()
 plt.show()

pyriemann_qiskit/__init__.py

@@ -1,4 +1,10 @@
 from ._version import __version__
 from . import classification, pipelines, ensemble, autoencoders
 
-__all__ = ["__version__", "classification", "pipelines", "ensemble", "autoencoders"]
+__all__ = [
+    "__version__",
+    "classification",
+    "pipelines",
+    "ensemble",
+    "autoencoders",
+]

pyriemann_qiskit/pipelines.py

@@ -285,7 +285,7 @@ def _create_pipe(self):
                 gen_feature_map=feature_map,
                 shots=self.shots,
                 quantum=is_quantum,
-                **self.params
+                **self.params,
             )
         else:
             self._log("QuanticSVM chosen.")
@@ -296,7 +296,7 @@ def _create_pipe(self):
                 max_iter=self.max_iter,
                 gen_feature_map=feature_map,
                 shots=self.shots,
-                **self.params
+                **self.params,
             )
 
         return make_pipeline(
@@ -402,8 +402,7 @@ def __init__(
         BasePipeline.__init__(self, "QuantumMDMWithRiemannianPipeline")
 
     def _create_pipe(self):
-        print(self.metric)
-        print(self.metric["mean"])
+        self._log(f"Running QMDM with metric {self.metric}")
         if is_qfunction(self.metric["mean"]):
             if self.quantum:
                 covariances = XdawnCovariances(