libmultilabel 0.7.0__tar.gz → 0.7.2__tar.gz

{libmultilabel-0.7.0 → libmultilabel-0.7.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: libmultilabel
-Version: 0.7.0
+Version: 0.7.2
 Summary: A library for multi-class and multi-label classification
 Home-page: https://github.com/ASUS-AICS/LibMultiLabel
 Author: LibMultiLabel Team
@@ -24,12 +24,12 @@ Requires-Dist: numba
 Requires-Dist: pandas>1.3.0
 Requires-Dist: PyYAML
 Requires-Dist: scikit-learn
-Requires-Dist: scipy
+Requires-Dist: scipy<1.14.0
 Requires-Dist: tqdm
 Provides-Extra: nn
 Requires-Dist: lightning==2.0.9; extra == "nn"
 Requires-Dist: nltk; extra == "nn"
-Requires-Dist: torch; extra == "nn"
+Requires-Dist: torch<=2.3; extra == "nn"
 Requires-Dist: torchmetrics==0.10.3; extra == "nn"
 Requires-Dist: torchtext; extra == "nn"
 Requires-Dist: transformers; extra == "nn"

{libmultilabel-0.7.0 → libmultilabel-0.7.2}/libmultilabel/linear/linear.py

@@ -79,7 +79,7 @@ def train_1vsrest(
     options: str = "",
     verbose: bool = True,
 ) -> FlatModel:
-    """Trains a linear model for multiabel data using a one-vs-rest strategy.
+    """Trains a linear model for multi-label data using a one-vs-rest strategy.
 
     Args:
         y (sparse.csr_matrix): A 0/1 matrix with dimensions number of instances * number of classes.
@@ -139,7 +139,7 @@ def _prepare_options(x: sparse.csr_matrix, options: str) -> tuple[sparse.csr_mat
             raise ValueError("Invalid LIBLINEAR solver type. Only classification solvers are allowed.")
     else:
         # workaround for liblinear warning about unspecified solver
-        options_split.extend(["-s", "2"])
+        options_split.extend(["-s", "1"])
 
     bias = -1.0
     if "-B" in options_split:
@@ -396,7 +396,7 @@ def train_cost_sensitive(
     options: str = "",
     verbose: bool = True,
 ) -> FlatModel:
-    """Trains a linear model for multilabel data using a one-vs-rest strategy
+    """Trains a linear model for multi-label data using a one-vs-rest strategy
     and cross-validation to pick an optimal asymmetric misclassification cost
     for Macro-F1.
     Outperforms train_1vsrest in most aspects at the cost of higher
@@ -500,7 +500,7 @@ def train_cost_sensitive_micro(
     options: str = "",
     verbose: bool = True,
 ) -> FlatModel:
-    """Trains a linear model for multilabel data using a one-vs-rest strategy
+    """Trains a linear model for multi-label data using a one-vs-rest strategy
     and cross-validation to pick an optimal asymmetric misclassification cost
     for Micro-F1.
     Outperforms train_1vsrest in most aspects at the cost of higher

{libmultilabel-0.7.0 → libmultilabel-0.7.2}/libmultilabel/linear/metrics.py

@@ -300,13 +300,13 @@ def get_metrics(monitor_metrics: list[str], num_classes: int, multiclass: bool =
         monitor_metrics = []
     metrics = {}
     for metric in monitor_metrics:
-        if re.match("P@\d+", metric):
+        if re.match(r"P@\d+", metric):
             metrics[metric] = PrecisionAtK(top_k=int(metric[2:]))
-        elif re.match("R@\d+", metric):
+        elif re.match(r"R@\d+", metric):
             metrics[metric] = RecallAtK(top_k=int(metric[2:]))
-        elif re.match("RP@\d+", metric):
+        elif re.match(r"RP@\d+", metric):
             metrics[metric] = RPrecisionAtK(top_k=int(metric[3:]))
-        elif re.match("NDCG@\d+", metric):
+        elif re.match(r"NDCG@\d+", metric):
             metrics[metric] = NDCGAtK(top_k=int(metric[5:]))
         elif metric in {"Another-Macro-F1", "Macro-F1", "Micro-F1"}:
             metrics[metric] = F1(num_classes, average=metric[:-3].lower(), multiclass=multiclass)

{libmultilabel-0.7.0 → libmultilabel-0.7.2}/libmultilabel/linear/tree.py

@@ -7,6 +7,7 @@ import scipy.sparse as sparse
 import sklearn.cluster
 import sklearn.preprocessing
 from tqdm import tqdm
+import psutil
 
 from . import linear
 
@@ -26,6 +27,7 @@ class Node:
         """
         self.label_map = label_map
         self.children = children
+        self.is_root = False
 
     def isLeaf(self) -> bool:
         return len(self.children) == 0
@@ -57,7 +59,7 @@ class TreeModel:
         x: sparse.csr_matrix,
         beam_width: int = 10,
     ) -> np.ndarray:
-        """Calculates the decision values associated with x.
+        """Calculates the probability estimates associated with x.
 
         Args:
             x (sparse.csr_matrix): A matrix with dimension number of instances * number of features.
@@ -71,10 +73,10 @@ class TreeModel:
         return np.vstack([self._beam_search(all_preds[i], beam_width) for i in range(all_preds.shape[0])])
 
     def _beam_search(self, instance_preds: np.ndarray, beam_width: int) -> np.ndarray:
-        """Predict with beam search using cached decision values for a single instance.
+        """Predict with beam search using cached probability estimates for a single instance.
 
         Args:
-            instance_preds (np.ndarray): A vector of cached decision values of each node, has dimension number of labels + total number of metalabels.
+            instance_preds (np.ndarray): A vector of cached probability estimates of each node, has dimension number of labels + total number of metalabels.
             beam_width (int): Number of candidates considered.
 
         Returns:
@@ -93,18 +95,18 @@ class TreeModel:
                     continue
                 slice = np.s_[self.weight_map[node.index] : self.weight_map[node.index + 1]]
                 pred = instance_preds[slice]
-                children_score = score - np.maximum(0, 1 - pred) ** 2
+                children_score = score - np.square(np.maximum(0, 1 - pred))
                 next_level.extend(zip(node.children, children_score.tolist()))
 
             cur_level = sorted(next_level, key=lambda pair: -pair[1])[:beam_width]
             next_level = []
 
         num_labels = len(self.root.label_map)
-        scores = np.full(num_labels, -np.inf)
+        scores = np.full(num_labels, 0.0)
         for node, score in cur_level:
             slice = np.s_[self.weight_map[node.index] : self.weight_map[node.index + 1]]
             pred = instance_preds[slice]
-            scores[node.label_map] = np.exp(score - np.maximum(0, 1 - pred) ** 2)
+            scores[node.label_map] = np.exp(score - np.square(np.maximum(0, 1 - pred)))
         return scores
 
 
@@ -116,7 +118,7 @@ def train_tree(
     dmax=10,
     verbose: bool = True,
 ) -> TreeModel:
-    """Trains a linear model for multiabel data using a divide-and-conquer strategy.
+    """Trains a linear model for multi-label data using a divide-and-conquer strategy.
     The algorithm used is based on https://github.com/xmc-aalto/bonsai.
 
     Args:
@@ -133,20 +135,39 @@ def train_tree(
     label_representation = (y.T * x).tocsr()
     label_representation = sklearn.preprocessing.normalize(label_representation, norm="l2", axis=1)
     root = _build_tree(label_representation, np.arange(y.shape[1]), 0, K, dmax)
+    root.is_root = True
 
     num_nodes = 0
+    # Both type(x) and type(y) are sparse.csr_matrix
+    # However, type((x != 0).T) becomes sparse.csc_matrix
+    # So type((x != 0).T * y) results in sparse.csc_matrix
+    features_used_perlabel = (x != 0).T * y
 
     def count(node):
         nonlocal num_nodes
         num_nodes += 1
+        node.num_features_used = np.count_nonzero(features_used_perlabel[:, node.label_map].sum(axis=1))
 
     root.dfs(count)
 
+    model_size = get_estimated_model_size(root)
+    print(f"The estimated tree model size is: {model_size / (1024**3):.3f} GB")
+
+    # Calculate the total memory (excluding swap) on the local machine
+    total_memory = psutil.virtual_memory().total
+    print(f"Your system memory is: {total_memory / (1024**3):.3f} GB")
+
+    if total_memory <= model_size:
+        raise MemoryError(f"Not enough memory to train the model.")
+
     pbar = tqdm(total=num_nodes, disable=not verbose)
 
     def visit(node):
-        relevant_instances = y[:, node.label_map].getnnz(axis=1) > 0
-        _train_node(y[relevant_instances], x[relevant_instances], options, node)
+        if node.is_root:
+            _train_node(y, x, options, node)
+        else:
+            relevant_instances = y[:, node.label_map].getnnz(axis=1) > 0
+            _train_node(y[relevant_instances], x[relevant_instances], options, node)
         pbar.update()
 
     root.dfs(visit)
@@ -195,6 +216,24 @@ def _build_tree(label_representation: sparse.csr_matrix, label_map: np.ndarray,
     return Node(label_map=label_map, children=children)
 
 
+def get_estimated_model_size(root):
+    total_num_weights = 0
+
+    def collect_stat(node: Node):
+        nonlocal total_num_weights
+
+        if node.isLeaf():
+            total_num_weights += len(node.label_map) * node.num_features_used
+        else:
+            total_num_weights += len(node.children) * node.num_features_used
+
+    root.dfs(collect_stat)
+
+    # 16 is because when storing sparse matrices, indices (int64) require 8 bytes and floats require 8 bytes
+    # Our study showed that among the used features of every binary classification problem, on average no more than 2/3 of weights obtained by the dual coordinate descent method are non-zeros.
+    return total_num_weights * 16 * 2 / 3
+
+
 def _train_node(y: sparse.csr_matrix, x: sparse.csr_matrix, options: str, node: Node):
     """If node is internal, computes the metalabels representing each child and trains
     on the metalabels. Otherwise, train on y.

{libmultilabel-0.7.0 → libmultilabel-0.7.2}/libmultilabel/linear/utils.py

@@ -76,17 +76,18 @@ class MultiLabelEstimator(sklearn.base.BaseEstimator):
         scoring_metric (str, optional): The scoring metric. Defaults to 'P@1'.
     """
 
-    def __init__(self, options: str = "", linear_technique: str = "1vsrest", scoring_metric: str = "P@1"):
+    def __init__(self, options: str = "", linear_technique: str = "1vsrest", scoring_metric: str = "P@1", multiclass: bool = False):
         super().__init__()
         self.options = options
         self.linear_technique = linear_technique
         self.scoring_metric = scoring_metric
         self._is_fitted = False
+        self.multiclass = multiclass
 
     def fit(self, X: sparse.csr_matrix, y: sparse.csr_matrix):
         X, y = sklearn.utils.validation.check_X_y(X, y, accept_sparse=True, multi_output=True)
         self._is_fitted = True
-        self.model = LINEAR_TECHNIQUES[self.linear_technique](y, X, self.options)
+        self.model = LINEAR_TECHNIQUES[self.linear_technique](y, X, options=self.options)
         return self
 
     def predict(self, X: sparse.csr_matrix) -> np.ndarray:
@@ -96,8 +97,9 @@ class MultiLabelEstimator(sklearn.base.BaseEstimator):
 
     def score(self, X: sparse.csr_matrix, y: sparse.csr_matrix) -> float:
         metrics = linear.get_metrics(
-            [self.scoring_metric],
-            y.shape[1],
+            monitor_metrics=[self.scoring_metric],
+            num_classes=y.shape[1],
+            multiclass=self.multiclass
         )
         preds = self.predict(X)
         metrics.update(preds, y.toarray())

{libmultilabel-0.7.0 → libmultilabel-0.7.2}/libmultilabel.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: libmultilabel
-Version: 0.7.0
+Version: 0.7.2
 Summary: A library for multi-class and multi-label classification
 Home-page: https://github.com/ASUS-AICS/LibMultiLabel
 Author: LibMultiLabel Team
@@ -24,12 +24,12 @@ Requires-Dist: numba
 Requires-Dist: pandas>1.3.0
 Requires-Dist: PyYAML
 Requires-Dist: scikit-learn
-Requires-Dist: scipy
+Requires-Dist: scipy<1.14.0
 Requires-Dist: tqdm
 Provides-Extra: nn
 Requires-Dist: lightning==2.0.9; extra == "nn"
 Requires-Dist: nltk; extra == "nn"
-Requires-Dist: torch; extra == "nn"
+Requires-Dist: torch<=2.3; extra == "nn"
 Requires-Dist: torchmetrics==0.10.3; extra == "nn"
 Requires-Dist: torchtext; extra == "nn"
 Requires-Dist: transformers; extra == "nn"

{libmultilabel-0.7.0 → libmultilabel-0.7.2}/libmultilabel.egg-info/requires.txt

@@ -3,13 +3,13 @@ numba
 pandas>1.3.0
 PyYAML
 scikit-learn
-scipy
+scipy<1.14.0
 tqdm
 
 [nn]
 lightning==2.0.9
 nltk
-torch
+torch<=2.3
 torchmetrics==0.10.3
 torchtext
 transformers

{libmultilabel-0.7.0 → libmultilabel-0.7.2}/setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = libmultilabel
-version = 0.7.0
+version = 0.7.2
 author = LibMultiLabel Team
 license = MIT License
 license_file = LICENSE
@@ -30,7 +30,7 @@ install_requires =
 	pandas>1.3.0
 	PyYAML
 	scikit-learn
-	scipy
+	scipy<1.14.0
 	tqdm
 python_requires = >=3.8
 
@@ -38,7 +38,7 @@ python_requires = >=3.8
 nn = 
 	lightning==2.0.9
 	nltk
-	torch
+	torch<=2.3
 	torchmetrics==0.10.3
 	torchtext
 	transformers