classifier updates

Author: dyumanaditya

examples/classifier_integration_ex.py

@@ -2,6 +2,15 @@
 import torch
 import torch.nn as nn
 import networkx as nx
+import numpy as np
+import random
+
+# seed_value = 41
+seed_value = 42
+random.seed(seed_value)
+np.random.seed(seed_value)
+torch.manual_seed(seed_value)
+
 
 # --- Part 1: Fraud Detector Model Integration ---
 
@@ -13,20 +22,20 @@
 # Create a dummy transaction feature vector.
 transaction_features = torch.rand(1, 5)
 
-# Define integration options.
+# Define integration options
 # Only probabilities above 0.5 are considered for adjustment.
 interface_options = pr.ModelInterfaceOptions(
-    threshold=0.5,       # Only process probabilities above 0.5
+    threshold=0.5,         # Only process probabilities above 0.5
     set_lower_bound=True,  # For high confidence, adjust the lower bound.
     set_upper_bound=False, # Keep the upper bound unchanged.
-    snap_value=1.0      # Use 1.0 as the snap value.
+    snap_value=1.0         # Use 1.0 as the snap value.
 )
 
 # Wrap the model using LogicIntegratedClassifier
 fraud_detector = pr.LogicIntegratedClassifier(
     model,
     class_names,
-    model_name="fraud_detector",
+    identifier="fraud_detector",
     interface_options=interface_options
 )
 
@@ -65,7 +74,7 @@
 
 # Define a rule: if the fraud detector flags a transaction as fraud and the transaction info is suspicious,
 # then mark the associated account (AccountA) as requiring investigation.
-pr.add_rule(pr.Rule("requires_investigation(acc) <- account(acc), fraud_detector(fraud), suspicious_location(acc)", "investigation_rule"))
+pr.add_rule(pr.Rule("requires_investigation(acc) <- account(acc), fraud(fraud_detector), suspicious_location(acc)", "investigation_rule"))
 
 # Define a propagation rule:
 # If an account requires investigation and is connected (via the "associated" relationship) to another account,
@@ -75,7 +84,7 @@
 # --- Part 4: Run the Reasoning Engine ---
 
 # Run the reasoning engine to allow the investigation flag to propagate through the network.
-pr.settings.allow_ground_rules = True
+# pr.settings.allow_ground_rules = True
 pr.settings.atom_trace = True
 interpretation = pr.reason()
 

pyreason/scripts/learning/classification/classifier.py

@@ -12,15 +12,17 @@ class LogicIntegratedClassifier(torch.nn.Module):
     Class to integrate a PyTorch model with PyReason. The output of the model is returned to the
     user in the form of PyReason facts. The user can then add these facts to the logic program and reason using them.
     """
-    def __init__(self, model, class_names: List[str], model_name: str = 'classifier', interface_options: ModelInterfaceOptions = None):
+    def __init__(self, model, class_names: List[str], identifier: str = 'classifier', interface_options: ModelInterfaceOptions = None):
         """
-        :param model:
-        :param class_names:
+        :param model: PyTorch model to be integrated.
+        :param class_names: List of class names for the model output.
+        :param identifier: Identifier for the model, used as the constant in the facts.
+        :param interface_options: Options for the model interface, including threshold and snapping behavior.
         """
         super(LogicIntegratedClassifier, self).__init__()
         self.model = model
         self.class_names = class_names
-        self.model_name = model_name
+        self.identifier = identifier
         self.interface_options = interface_options
 
     def get_class_facts(self, t1: int, t2: int) -> List[Fact]:
@@ -33,7 +35,7 @@ def get_class_facts(self, t1: int, t2: int) -> List[Fact]:
         """
         facts = []
         for c in self.class_names:
-            fact = Fact(f'{self.model_name}({c})', name=f'{self.model_name}-{c}-fact', start_time=t1, end_time=t2)
+            fact = Fact(f'{c}({self.identifier})', name=f'{self.identifier}-{c}-fact', start_time=t1, end_time=t2)
             facts.append(fact)
         return facts
 
@@ -82,8 +84,8 @@ def forward(self, x, t1: int = 0, t2: int = 0) -> Tuple[torch.Tensor, torch.Tens
         facts = []
         for class_name, bounds in zip(self.class_names, bounds_list):
             lower, upper = bounds
-            fact_str = f'{self.model_name}({class_name}) : [{lower:.3f}, {upper:.3f}]'
-            fact = Fact(fact_str, name=f'{self.model_name}-{class_name}-fact', start_time=t1, end_time=t2)
+            fact_str = f'{class_name}({self.identifier}) : [{lower:.3f}, {upper:.3f}]'
+            fact = Fact(fact_str, name=f'{self.identifier}-{class_name}-fact', start_time=t1, end_time=t2)
             facts.append(fact)
         return output, probabilities, facts