Changeset 15739 for trunk/HeuristicLab.Algorithms.DataAnalysis/3.4/NeuralNetwork/NeuralNetworkEnsembleModel.cs

Timestamp:

02/08/18 14:16:39 (6 years ago)

Author:

gkronber

Message:

#2891: fixed bug in NN and NN ensemble using a simple lock for processing an input.

File:

• trunk/HeuristicLab.Algorithms.DataAnalysis/3.4/NeuralNetwork/NeuralNetworkEnsembleModel.cs (modified) (2 diffs)

trunk/HeuristicLab.Algorithms.DataAnalysis/3.4/NeuralNetwork/NeuralNetworkEnsembleModel.cs

@@ -102 +102 @@
           x[column] = inputData[row, column];
         }
-        alglib.mlpeprocess(mlpEnsemble, x, ref y);
+        // mlpeprocess writes data in mlpEnsemble and is therefore not thread-safe
+        lock (mlpEnsemble) {
+          alglib.mlpeprocess(mlpEnsemble, x, ref y);
+        }
         yield return y[0];
       }
@@ -119 +122 @@
           x[column] = inputData[row, column];
         }
-        alglib.mlpeprocess(mlpEnsemble, x, ref y);
+        // mlpeprocess writes data in mlpEnsemble and is therefore not thread-safe
+        lock (mlpEnsemble) {
+          alglib.mlpeprocess(mlpEnsemble, x, ref y);
+        }
         // find class for with the largest probability value
         int maxProbClassIndex = 0;