add XGBoost Image Classifier test function

Author: SimonBlanke

src/surfaces/test_functions/machine_learning/image/__init__.py

@@ -15,6 +15,14 @@
 except ImportError:
     _HAS_TENSORFLOW = False
 
+# XGBoost classifier (requires xgboost)
+try:
+    from .classification import XGBoostImageClassifierFunction
+
+    _HAS_XGBOOST = True
+except ImportError:
+    _HAS_XGBOOST = False
+
 __all__ = [
     "SVMImageClassifierFunction",
     "RandomForestImageClassifierFunction",
@@ -27,3 +35,6 @@
             "DeepCNNClassifierFunction",
         ]
     )
+
+if _HAS_XGBOOST:
+    __all__.append("XGBoostImageClassifierFunction")

src/surfaces/test_functions/machine_learning/image/classification/__init__.py

@@ -15,6 +15,14 @@
 except ImportError:
     _HAS_TENSORFLOW = False
 
+# XGBoost classifier (requires xgboost)
+try:
+    from .test_functions import XGBoostImageClassifierFunction
+
+    _HAS_XGBOOST = True
+except ImportError:
+    _HAS_XGBOOST = False
+
 __all__ = [
     "SVMImageClassifierFunction",
     "RandomForestImageClassifierFunction",
@@ -27,3 +35,6 @@
             "DeepCNNClassifierFunction",
         ]
     )
+
+if _HAS_XGBOOST:
+    __all__.append("XGBoostImageClassifierFunction")

src/surfaces/test_functions/machine_learning/image/classification/test_functions/__init__.py

@@ -14,6 +14,14 @@
 except ImportError:
     _HAS_TENSORFLOW = False
 
+# XGBoost classifier (requires xgboost)
+try:
+    from .xgboost_image_classifier import XGBoostImageClassifierFunction
+
+    _HAS_XGBOOST = True
+except ImportError:
+    _HAS_XGBOOST = False
+
 __all__ = [
     "SVMImageClassifierFunction",
     "RandomForestImageClassifierFunction",
@@ -26,3 +34,6 @@
             "DeepCNNClassifierFunction",
         ]
     )
+
+if _HAS_XGBOOST:
+    __all__.append("XGBoostImageClassifierFunction")

src/surfaces/test_functions/machine_learning/image/classification/test_functions/xgboost_image_classifier.py

@@ -0,0 +1,157 @@
+# Author: Simon Blanke
+# Email: [email protected]
+# License: MIT License
+
+"""XGBoost Image Classifier test function."""
+
+import numpy as np
+
+from .._base_image_classification import BaseImageClassification
+from ..datasets import DATASETS
+
+
+def _check_xgboost():
+    """Check if xgboost is available."""
+    try:
+        import xgboost  # noqa: F401
+
+        return True
+    except ImportError:
+        raise ImportError(
+            "XGBoost image classifier requires xgboost. "
+            "Install with: pip install surfaces[xgboost]"
+        )
+
+
+class XGBoostImageClassifierFunction(BaseImageClassification):
+    """XGBoost Image Classifier test function.
+
+    Uses XGBoost on PCA-reduced image features for classification.
+    XGBoost is a gradient boosting library optimized for speed and performance.
+
+    Parameters
+    ----------
+    dataset : str, default="mnist"
+        Dataset to use. One of: "mnist", "fashion_mnist".
+    cv : int, default=3
+        Number of cross-validation folds.
+    n_components : int, default=50
+        Number of PCA components to retain.
+    use_surrogate : bool, default=False
+        If True, use pre-trained surrogate for fast evaluation.
+    objective : str, default="maximize"
+        Either "minimize" or "maximize".
+    sleep : float, default=0
+        Artificial delay in seconds.
+
+    Examples
+    --------
+    >>> from surfaces.test_functions.machine_learning.image import (
+    ...     XGBoostImageClassifierFunction
+    ... )
+    >>> func = XGBoostImageClassifierFunction(dataset="mnist")
+    >>> result = func({"n_estimators": 100, "max_depth": 6, "learning_rate": 0.1})
+    """
+
+    name = "XGBoost Image Classifier Function"
+    _name_ = "xgboost_image_classifier"
+    __name__ = "XGBoostImageClassifierFunction"
+
+    available_datasets = list(DATASETS.keys())
+    available_cv = [2, 3, 5]
+
+    # Search space parameters
+    para_names = ["n_estimators", "max_depth", "learning_rate"]
+    n_estimators_default = [50, 100, 150, 200, 250]
+    max_depth_default = [3, 4, 5, 6, 7, 8, 10]
+    learning_rate_default = [0.01, 0.05, 0.1, 0.2, 0.3]
+
+    def __init__(
+        self,
+        dataset: str = "mnist",
+        cv: int = 3,
+        n_components: int = 50,
+        objective: str = "maximize",
+        sleep: float = 0,
+        memory: bool = False,
+        collect_data: bool = True,
+        callbacks=None,
+        catch_errors=None,
+        use_surrogate: bool = False,
+    ):
+        _check_xgboost()
+
+        if dataset not in DATASETS:
+            raise ValueError(
+                f"Unknown dataset '{dataset}'. " f"Available: {self.available_datasets}"
+            )
+
+        if cv not in self.available_cv:
+            raise ValueError(f"Invalid cv={cv}. Available: {self.available_cv}")
+
+        self.dataset = dataset
+        self.cv = cv
+        self.n_components = n_components
+        self._dataset_loader = DATASETS[dataset]
+
+        super().__init__(
+            objective=objective,
+            sleep=sleep,
+            memory=memory,
+            collect_data=collect_data,
+            callbacks=callbacks,
+            catch_errors=catch_errors,
+            use_surrogate=use_surrogate,
+        )
+
+    @property
+    def search_space(self):
+        """Search space containing hyperparameters."""
+        return {
+            "n_estimators": self.n_estimators_default,
+            "max_depth": self.max_depth_default,
+            "learning_rate": self.learning_rate_default,
+        }
+
+    def _create_objective_function(self):
+        """Create objective function with fixed dataset and cv."""
+        from sklearn.decomposition import PCA
+        from sklearn.model_selection import cross_val_score
+        from sklearn.preprocessing import StandardScaler
+        from xgboost import XGBClassifier
+
+        X_raw, y = self._dataset_loader()
+
+        # Apply PCA for dimensionality reduction
+        scaler = StandardScaler()
+        pca = PCA(n_components=self.n_components, random_state=42)
+        X_scaled = scaler.fit_transform(X_raw)
+        X = pca.fit_transform(X_scaled)
+
+        cv = self.cv
+        n_classes = len(np.unique(y))
+
+        def xgboost_image_classifier(params):
+            model = XGBClassifier(
+                n_estimators=params["n_estimators"],
+                max_depth=params["max_depth"],
+                learning_rate=params["learning_rate"],
+                objective="multi:softmax" if n_classes > 2 else "binary:logistic",
+                num_class=n_classes if n_classes > 2 else None,
+                random_state=42,
+                n_jobs=-1,
+                verbosity=0,
+            )
+            scores = cross_val_score(model, X, y, cv=cv, scoring="accuracy")
+            return scores.mean()
+
+        self.pure_objective_function = xgboost_image_classifier
+
+    def _get_surrogate_params(self, params):
+        """Add fixed parameters for surrogate prediction."""
+        return {
+            **params,
+            "dataset": self.dataset,
+            "cv": self.cv,
+            "n_components": self.n_components,
+        }