diff --git a/metric_learn/_util.py b/metric_learn/_util.py
index 868ececa..de608218 100644
--- a/metric_learn/_util.py
+++ b/metric_learn/_util.py
@@ -1,5 +1,6 @@
 import numpy as np
 from numpy.linalg import LinAlgError
+from inspect import signature
 from sklearn.datasets import make_spd_matrix
 from sklearn.decomposition import PCA
 from sklearn.utils import check_array
@@ -22,6 +23,29 @@ def vector_norm(X):
     return np.linalg.norm(X, axis=1)
 
 
+_CHECK_ARRAY_SUPPORTS_FORCE_ALL_FINITE = (
+    'force_all_finite' in signature(check_array).parameters)
+_CHECK_X_Y_SUPPORTS_FORCE_ALL_FINITE = (
+    'force_all_finite' in signature(check_X_y).parameters)
+
+
+def _check_array(*args, **kwargs):
+  """Local wrapper around `sklearn.utils.check_array` to deal with the change
+  from `force_all_finite` to `ensure_all_finite` in scikit-learn."""
+  if not _CHECK_ARRAY_SUPPORTS_FORCE_ALL_FINITE and "force_all_finite" in kwargs:
+    kwargs = kwargs.copy()
+    kwargs["ensure_all_finite"] = kwargs.pop("force_all_finite")
+  return check_array(*args, **kwargs)
+
+def _check_X_y(*args, **kwargs):
+  """Local wrapper around `sklearn.utils.check_X_y` to deal with the change
+  from `force_all_finite` to `ensure_all_finite` in scikit-learn."""
+  if not _CHECK_X_Y_SUPPORTS_FORCE_ALL_FINITE and "force_all_finite" in kwargs:
+    kwargs = kwargs.copy()
+    kwargs["ensure_all_finite"] = kwargs.pop("force_all_finite")
+  return check_X_y(*args, **kwargs)
+
+
 def check_input(input_data, y=None, preprocessor=None,
                 type_of_inputs='classic', tuple_size=None, accept_sparse=False,
                 dtype='numeric', order=None,
@@ -115,14 +139,14 @@ def check_input(input_data, y=None, preprocessor=None,
 
   # We need to convert input_data into a numpy.ndarray if possible, before
   # any further checks or conversions, and deal with y if needed. Therefore
-  # we use check_array/check_X_y with fixed permissive arguments.
+  # we use the wrappers _check_array/_check_X_y with fixed permissive arguments.
   if y is None:
-    input_data = check_array(input_data, ensure_2d=False, allow_nd=True,
+    input_data = _check_array(input_data, ensure_2d=False, allow_nd=True,
                              copy=False, force_all_finite=False,
                              accept_sparse=True, dtype=None,
                              ensure_min_features=0, ensure_min_samples=0)
   else:
-    input_data, y = check_X_y(input_data, y, ensure_2d=False, allow_nd=True,
+    input_data, y = _check_X_y(input_data, y, ensure_2d=False, allow_nd=True,
                               copy=False, force_all_finite=False,
                               accept_sparse=True, dtype=None,
                               ensure_min_features=0, ensure_min_samples=0,
@@ -165,9 +189,9 @@ def check_input_tuples(input_data, context, preprocessor, args_for_sk_checks,
       make_error_input(420, input_data, context)
     else:
       make_error_input(200, input_data, context)
-  input_data = check_array(input_data, allow_nd=True, ensure_2d=False,
+  input_data = _check_array(input_data, allow_nd=True, ensure_2d=False,
                            **args_for_sk_checks)
-  # we need to check num_features because check_array does not check it
+  # we need to check num_features because _check_array does not check it
   # for 3D inputs:
   if args_for_sk_checks['ensure_min_features'] > 0:
     n_features = input_data.shape[2]
@@ -180,7 +204,7 @@ def check_input_tuples(input_data, context, preprocessor, args_for_sk_checks,
   #  normally we don't need to check_tuple_size too because tuple_size
   # shouldn't be able to be modified by any preprocessor
   if input_data.ndim != 3:
-    # we have to ensure this because check_array above does not
+    # we have to ensure this because _check_array above does not
     if preprocessor_has_been_applied:
       make_error_input(211, input_data, context)
     else:
@@ -205,10 +229,10 @@ def check_input_classic(input_data, context, preprocessor, args_for_sk_checks):
     else:
       make_error_input(100, input_data, context)
 
-  input_data = check_array(input_data, allow_nd=True, ensure_2d=False,
+  input_data = _check_array(input_data, allow_nd=True, ensure_2d=False,
                            **args_for_sk_checks)
   if input_data.ndim != 2:
-    # we have to ensure this because check_array above does not
+    # we have to ensure this because _check_array above does not
     if preprocessor_has_been_applied:
       make_error_input(111, input_data, context)
     else:
@@ -317,7 +341,7 @@ def __init__(self, X):
     # format with arguments in check_input, and only this latter function
     # should return the appropriate errors). We do this only to have a numpy
     # array object which can be indexed by another numpy array object.
-    X = check_array(X,
+    X = _check_array(X,
                     accept_sparse=True, dtype=None,
                     force_all_finite=False,
                     ensure_2d=False, allow_nd=True,
@@ -514,7 +538,7 @@ def _initialize_components(n_components, input, y=None, init='auto',
   if isinstance(init, np.ndarray):
     # we copy the array, so that if we update the metric, we don't want to
     # update the init
-    init = check_array(init, copy=True)
+    init = _check_array(init, copy=True)
 
     # Assert that init.shape[1] = X.shape[1]
     if init.shape[1] != n_features:
@@ -656,7 +680,7 @@ def _initialize_metric_mahalanobis(input, init='identity', random_state=None,
   if isinstance(init, np.ndarray):
     # we copy the array, so that if we update the metric, we don't want to
     # update the init
-    init = check_array(init, copy=True)
+    init = _check_array(init, copy=True)
 
     # Assert that init.shape[1] = n_features
     if init.shape != (n_features,) * 2: