fix: compatibility with sklearn 1.4 and 1.1

src/concrete/ml/sklearn/base.py

@@ -1572,6 +1572,29 @@ def post_processing(self, y_preds: numpy.ndarray) -> numpy.ndarray:
 
         return super().post_processing(y_preds)
 
+    def get_sklearn_params(self, deep: bool = True) -> dict:
+        """Get parameters for this estimator.
+
+        This method is used to instantiate a scikit-learn model using the Concrete ML model's
+        parameters. It does not override scikit-learn's existing `get_params` method in order to
+        not break its implementation of `set_params`.
+
+        Args:
+            deep (bool): If True, will return the parameters for this estimator and contained
+                subobjects that are estimators. Default to True.
+
+        Returns:
+            params (dict): Parameter names mapped to their values.
+        """
+        # pylint: disable-next=no-member
+        params = super().get_params(deep=deep)  # type: ignore[misc]
+
+        params.pop("n_bits", None)
+        if "1.1." in sklearn.__version__:
+            params.pop("monotonic_cst", None)
+
+        return params
+
 
 class BaseTreeRegressorMixin(BaseTreeEstimatorMixin, sklearn.base.RegressorMixin, ABC):
     """Mixin class for tree-based regressors.
@@ -1668,6 +1691,12 @@ def from_sklearn_model(
         # Extract scikit-learn's initialization parameters
         init_params = sklearn_model.get_params()
 
+        # Ensure compatibility for both sklearn 1.1 and >=1.4
+        # This parameter was removed in 1.4. If this package is installed
+        # with sklearn 1.1 which has it, then remove it when
+        # instantiating the 1.4 API compatible Concrete ML model
+        init_params.pop("normalize", None)
+
         # Instantiate the Concrete ML model and update initialization parameters
         # This update is necessary as we currently store scikit-learn attributes in Concrete ML
         # classes during initialization (for example: link or power attributes in GLMs)

src/concrete/ml/sklearn/glm.py

@@ -6,6 +6,7 @@
 from typing import Any, Dict, Union
 
 import numpy
+import sklearn
 import sklearn.linear_model
 
 from ..common.debugging.custom_assert import assert_true
@@ -136,6 +137,19 @@ def load_dict(cls, metadata: Dict):
 
         return obj
 
+    def get_sklearn_params(self, deep: bool = True) -> dict:
+        # Here, the `get_params` method is the `BaseEstimator.get_params` method from scikit-learn
+        # FIXME: https://github.com/zama-ai/concrete-ml-internal/issues/3373
+        params = super().get_params(deep=deep)  # type: ignore[misc]
+
+        # Remove the parameters added by Concrete ML
+        params.pop("n_bits", None)
+        # Remove sklearn 1.4 parameter when using sklearn 1.1
+        if "1.1." in sklearn.__version__:
+            params.pop("solver", None)
+
+        return params
+
 
 class PoissonRegressor(_GeneralizedLinearRegressor):
     """A Poisson regression model with FHE.
@@ -156,12 +170,13 @@ class PoissonRegressor(_GeneralizedLinearRegressor):
     sklearn_model_class = sklearn.linear_model.PoissonRegressor
     _is_a_public_cml_model = True
 
-    def __init__(
+    def __init__(  # pylint: disable=useless-parent-delegation
         self,
         *,
         n_bits: Union[int, dict] = 8,
         alpha: float = 1.0,
         fit_intercept: bool = True,
+        solver="lbfgs",
         max_iter: int = 100,
         tol: float = 1e-4,
         warm_start: bool = False,
@@ -171,6 +186,7 @@ def __init__(
             n_bits=n_bits,
             alpha=alpha,
             fit_intercept=fit_intercept,
+            solver=solver,
             max_iter=max_iter,
             tol=tol,
             warm_start=warm_start,
@@ -200,12 +216,13 @@ class GammaRegressor(_GeneralizedLinearRegressor):
     sklearn_model_class = sklearn.linear_model.GammaRegressor
     _is_a_public_cml_model = True
 
-    def __init__(
+    def __init__(  # pylint: disable=useless-parent-delegation
         self,
         *,
         n_bits: Union[int, dict] = 8,
         alpha: float = 1.0,
         fit_intercept: bool = True,
+        solver="lbfgs",
         max_iter: int = 100,
         tol: float = 1e-4,
         warm_start: bool = False,
@@ -215,6 +232,7 @@ def __init__(
             n_bits=n_bits,
             alpha=alpha,
             fit_intercept=fit_intercept,
+            solver=solver,
             max_iter=max_iter,
             tol=tol,
             warm_start=warm_start,
@@ -245,14 +263,15 @@ class TweedieRegressor(_GeneralizedLinearRegressor):
     sklearn_model_class = sklearn.linear_model.TweedieRegressor
     _is_a_public_cml_model = True
 
-    def __init__(
+    def __init__(  # pylint: disable=too-many-arguments
         self,
         *,
         n_bits: Union[int, dict] = 8,
         power: float = 0.0,
         alpha: float = 1.0,
         fit_intercept: bool = True,
         link: str = "auto",
+        solver="lbfgs",
         max_iter: int = 100,
         tol: float = 1e-4,
         warm_start: bool = False,
@@ -262,6 +281,7 @@ def __init__(
             n_bits=n_bits,
             alpha=alpha,
             fit_intercept=fit_intercept,
+            solver=solver,
             max_iter=max_iter,
             tol=tol,
             warm_start=warm_start,

src/concrete/ml/sklearn/linear_model.py

@@ -51,7 +51,6 @@ def __init__(
         self,
         n_bits=8,
         fit_intercept=True,
-        normalize="deprecated",
         copy_X=True,
         n_jobs=None,
         positive=False,
@@ -60,7 +59,6 @@ def __init__(
         super().__init__(n_bits=n_bits)
 
         self.fit_intercept = fit_intercept
-        self.normalize = normalize
         self.copy_X = copy_X
         self.n_jobs = n_jobs
         self.positive = positive
@@ -85,7 +83,6 @@ def dump_dict(self) -> Dict[str, Any]:
 
         # scikit-learn
         metadata["fit_intercept"] = self.fit_intercept
-        metadata["normalize"] = self.normalize
         metadata["copy_X"] = self.copy_X
         metadata["n_jobs"] = self.n_jobs
         metadata["positive"] = self.positive
@@ -112,7 +109,6 @@ def load_dict(cls, metadata: Dict):
 
         # scikit-learn
         obj.fit_intercept = metadata["fit_intercept"]
-        obj.normalize = metadata["normalize"]
         obj.copy_X = metadata["copy_X"]
         obj.n_jobs = metadata["n_jobs"]
         obj.positive = metadata["positive"]
@@ -1302,7 +1298,6 @@ def __init__(
         alpha=1.0,
         l1_ratio=0.5,
         fit_intercept=True,
-        normalize="deprecated",
         precompute=False,
         max_iter=1000,
         copy_X=True,
@@ -1318,7 +1313,6 @@ def __init__(
         self.alpha = alpha
         self.l1_ratio = l1_ratio
         self.fit_intercept = fit_intercept
-        self.normalize = normalize
         self.copy_X = copy_X
         self.positive = positive
         self.precompute = precompute
@@ -1350,7 +1344,6 @@ def dump_dict(self) -> Dict[str, Any]:
         metadata["alpha"] = self.alpha
         metadata["l1_ratio"] = self.l1_ratio
         metadata["fit_intercept"] = self.fit_intercept
-        metadata["normalize"] = self.normalize
         metadata["copy_X"] = self.copy_X
         metadata["positive"] = self.positive
         metadata["precompute"] = self.precompute
@@ -1384,7 +1377,6 @@ def load_dict(cls, metadata: Dict):
         obj.alpha = metadata["alpha"]
         obj.l1_ratio = metadata["l1_ratio"]
         obj.fit_intercept = metadata["fit_intercept"]
-        obj.normalize = metadata["normalize"]
         obj.copy_X = metadata["copy_X"]
         obj.positive = metadata["positive"]
         obj.precompute = metadata["precompute"]
@@ -1422,7 +1414,6 @@ def __init__(
         n_bits=8,
         alpha: float = 1.0,
         fit_intercept=True,
-        normalize="deprecated",
         precompute=False,
         copy_X=True,
         max_iter=1000,
@@ -1437,7 +1428,6 @@ def __init__(
 
         self.alpha = alpha
         self.fit_intercept = fit_intercept
-        self.normalize = normalize
         self.copy_X = copy_X
         self.positive = positive
         self.max_iter = max_iter
@@ -1468,7 +1458,6 @@ def dump_dict(self) -> Dict[str, Any]:
         # scikit-learn
         metadata["alpha"] = self.alpha
         metadata["fit_intercept"] = self.fit_intercept
-        metadata["normalize"] = self.normalize
         metadata["copy_X"] = self.copy_X
         metadata["positive"] = self.positive
         metadata["max_iter"] = self.max_iter
@@ -1501,7 +1490,6 @@ def load_dict(cls, metadata: Dict):
         # scikit-learn
         obj.alpha = metadata["alpha"]
         obj.fit_intercept = metadata["fit_intercept"]
-        obj.normalize = metadata["normalize"]
         obj.copy_X = metadata["copy_X"]
         obj.positive = metadata["positive"]
         obj.max_iter = metadata["max_iter"]
@@ -1539,10 +1527,9 @@ def __init__(
         n_bits=8,
         alpha: float = 1.0,
         fit_intercept=True,
-        normalize="deprecated",
         copy_X=True,
         max_iter=None,
-        tol=0.001,
+        tol=0.0001,
         solver="auto",
         positive=False,
         random_state=None,
@@ -1552,7 +1539,6 @@ def __init__(
 
         self.alpha = alpha
         self.fit_intercept = fit_intercept
-        self.normalize = normalize
         self.copy_X = copy_X
         self.positive = positive
         self.max_iter = max_iter
@@ -1581,7 +1567,6 @@ def dump_dict(self) -> Dict[str, Any]:
         # scikit-learn
         metadata["alpha"] = self.alpha
         metadata["fit_intercept"] = self.fit_intercept
-        metadata["normalize"] = self.normalize
         metadata["copy_X"] = self.copy_X
         metadata["positive"] = self.positive
         metadata["max_iter"] = self.max_iter
@@ -1612,7 +1597,6 @@ def load_dict(cls, metadata: Dict):
         # scikit-learn
         obj.alpha = metadata["alpha"]
         obj.fit_intercept = metadata["fit_intercept"]
-        obj.normalize = metadata["normalize"]
         obj.copy_X = metadata["copy_X"]
         obj.positive = metadata["positive"]
         obj.max_iter = metadata["max_iter"]

src/concrete/ml/sklearn/rf.py

@@ -39,6 +39,7 @@ def __init__(
         class_weight=None,
         ccp_alpha=0.0,
         max_samples=None,
+        monotonic_cst=None,
     ):
         """Initialize the RandomForestClassifier.
 
@@ -65,6 +66,7 @@ def __init__(
         self.max_leaf_nodes = max_leaf_nodes
         self.min_impurity_decrease = min_impurity_decrease
         self.ccp_alpha = ccp_alpha
+        self.monotonic_cst = monotonic_cst
 
     def post_processing(self, y_preds: numpy.ndarray) -> numpy.ndarray:
         # Here, we want to use BaseTreeEstimatorMixin's `post-processing` method as
@@ -106,6 +108,7 @@ def dump_dict(self) -> Dict[str, Any]:
         metadata["max_leaf_nodes"] = self.max_leaf_nodes
         metadata["min_impurity_decrease"] = self.min_impurity_decrease
         metadata["ccp_alpha"] = self.ccp_alpha
+        metadata["monotonic_cst"] = self.monotonic_cst
 
         return metadata
 
@@ -151,6 +154,7 @@ def load_dict(cls, metadata: Dict):
         obj.max_leaf_nodes = metadata["max_leaf_nodes"]
         obj.min_impurity_decrease = metadata["min_impurity_decrease"]
         obj.ccp_alpha = metadata["ccp_alpha"]
+        obj.monotonic_cst = metadata["monotonic_cst"]
 
         return obj
 
@@ -184,6 +188,7 @@ def __init__(
         warm_start=False,
         ccp_alpha=0.0,
         max_samples=None,
+        monotonic_cst=None,
     ):
         """Initialize the RandomForestRegressor.
 
@@ -209,6 +214,7 @@ def __init__(
         self.max_leaf_nodes = max_leaf_nodes
         self.min_impurity_decrease = min_impurity_decrease
         self.ccp_alpha = ccp_alpha
+        self.monotonic_cst = monotonic_cst
 
     def dump_dict(self) -> Dict[str, Any]:
         metadata: Dict[str, Any] = {}
@@ -243,6 +249,7 @@ def dump_dict(self) -> Dict[str, Any]:
         metadata["max_leaf_nodes"] = self.max_leaf_nodes
         metadata["min_impurity_decrease"] = self.min_impurity_decrease
         metadata["ccp_alpha"] = self.ccp_alpha
+        metadata["monotonic_cst"] = self.monotonic_cst
 
         return metadata
 
@@ -288,5 +295,6 @@ def load_dict(cls, metadata: Dict):
         obj.max_leaf_nodes = metadata["max_leaf_nodes"]
         obj.min_impurity_decrease = metadata["min_impurity_decrease"]
         obj.ccp_alpha = metadata["ccp_alpha"]
+        obj.monotonic_cst = metadata["monotonic_cst"]
 
         return obj

src/concrete/ml/sklearn/tree.py

@@ -32,6 +32,7 @@ def __init__(
         min_impurity_decrease=0.0,
         class_weight=None,
         ccp_alpha: float = 0.0,
+        monotonic_cst=None,
         n_bits: Union[int, Dict[str, int]] = 6,
     ):
         """Initialize the DecisionTreeClassifier.
@@ -54,6 +55,7 @@ def __init__(
         self.random_state = random_state
         self.min_impurity_decrease = min_impurity_decrease
         self.ccp_alpha = ccp_alpha
+        self.monotonic_cst = monotonic_cst
 
     def __getattr__(self, attr: str):
         # We directly expose the following methods as they are commonly used with decision trees
@@ -100,6 +102,7 @@ def dump_dict(self) -> Dict[str, Any]:
         metadata["random_state"] = self.random_state
         metadata["min_impurity_decrease"] = self.min_impurity_decrease
         metadata["ccp_alpha"] = self.ccp_alpha
+        metadata["monotonic_cst"] = self.monotonic_cst
 
         return metadata
 
@@ -140,6 +143,7 @@ def load_dict(cls, metadata: Dict):
         obj.random_state = metadata["random_state"]
         obj.min_impurity_decrease = metadata["min_impurity_decrease"]
         obj.ccp_alpha = metadata["ccp_alpha"]
+        obj.monotonic_cst = metadata["monotonic_cst"]
 
         return obj
 
@@ -166,6 +170,7 @@ def __init__(
         max_leaf_nodes=None,
         min_impurity_decrease=0.0,
         ccp_alpha=0.0,
+        monotonic_cst=None,
         n_bits: Union[int, Dict[str, int]] = 6,
     ):
         """Initialize the DecisionTreeRegressor.
@@ -187,6 +192,7 @@ def __init__(
         self.random_state = random_state
         self.min_impurity_decrease = min_impurity_decrease
         self.ccp_alpha = ccp_alpha
+        self.monotonic_cst = monotonic_cst
 
     def __getattr__(self, attr: str):
         # We directly expose the following methods as they are commonly used with decision trees
@@ -226,6 +232,7 @@ def dump_dict(self) -> Dict[str, Any]:
         metadata["random_state"] = self.random_state
         metadata["min_impurity_decrease"] = self.min_impurity_decrease
         metadata["ccp_alpha"] = self.ccp_alpha
+        metadata["monotonic_cst"] = self.monotonic_cst
 
         return metadata
 
@@ -266,5 +273,6 @@ def load_dict(cls, metadata: Dict):
         obj.random_state = metadata["random_state"]
         obj.min_impurity_decrease = metadata["min_impurity_decrease"]
         obj.ccp_alpha = metadata["ccp_alpha"]
+        obj.monotonic_cst = metadata["monotonic_cst"]
 
         return obj