concepts: add the base class of all concept methods

Signed-off-by: Frederic Boisnard <[email protected]>

xplique/concepts/base.py

@@ -0,0 +1,104 @@
+"""
+Module related to abstract concept explainer
+"""
+
+from abc import ABC, abstractmethod
+
+from sklearn.exceptions import NotFittedError
+
+from ..types import Callable, Optional
+
+class BaseConceptExtractor(ABC):
+
+    """
+    Base class for concept extraction models.
+
+    Parameters
+    ----------
+    number_of_concepts
+        The number of concepts to extract. Default is 20.
+    batch_size
+        The batch size to use during training and prediction. Default is 64.
+
+    """
+    @abstractmethod
+    def __init__(self, number_of_concepts: int = 20,
+                       batch_size: int = 64):
+        self.number_of_concepts = number_of_concepts
+        self.batch_size = batch_size
+
+        # sanity checks
+        assert(number_of_concepts > 0), "number_of_concepts must be greater than 0"
+        assert(batch_size > 0), "batch_size must be greater than 0"
+
+    @abstractmethod
+    def fit(self, inputs):
+        """
+        Fit the CAVs to the input data.
+
+        Parameters
+        ----------
+        inputs
+            The input data to fit the model on.
+
+        Returns
+        -------
+        tuple
+            A tuple containing the input data and the matrices (U, W) that factorize the data.
+
+        """
+        raise NotImplementedError
+
+    def check_if_fitted(self):
+        """Checks if the factorization model has been fitted to input data.
+
+        Raises
+        ------
+        NotFittedError
+            If the factorization model has not been fitted to input data.
+        """
+
+        if not hasattr(self, 'reducer'):
+            raise NotFittedError("The factorization model has not been fitted to input data yet.")
+
+    @abstractmethod
+    def transform(self, inputs):
+        """
+        Transform the input data into a concepts embedding.
+
+        Parameters
+        ----------
+        inputs
+            The input data to transform.
+
+        Returns
+        -------
+        array-like
+            The transformed embedding of the input data.
+
+        """
+        raise NotImplementedError
+
+
+class ClassifierConceptExtractor(BaseConceptExtractor):
+    """
+    Base class for concept extraction based on a single classifier model.
+
+    Parameters
+    ----------
+    model_encoder
+        The model to explain.
+    number_of_concepts
+        The number of concepts to extract. Default is 20.
+    batch_size
+        The batch size to use during training and prediction. Default is 64.
+
+    """
+    def __init__(self, model_encoder : Callable, 
+                       number_of_concepts: int = 20,
+                       batch_size: int = 64):
+        super().__init__(number_of_concepts, batch_size)
+        self.model_encoder = model_encoder
+
+        # sanity checks
+        assert(callable(model_encoder)), "model_encoder must be a callable function"