Error: "Training process was terminated unexpectedly" Vector Machine Learning (VML) Training Error

Products

• Data Loss Prevention Enforce
• Data Loss Prevention

Issue/Introduction

Attempts to configure the VML profile using a positive and negative set of data fail with the following error.

TRAINING ERROR 

Training process was terminated unexpectedly. Please check the logs for more details. 

Jul 17, 2014 2:46:09 PM com.vontu.machinelearning.training.MLDTrainingProcess exitWithError

INFO: Unhandeled exception in training process.

com.vontu.cracker.jni.NativeException: Failed to start Engine

at com.vontu.cracker.jni.NativeContentExtractionEngine.create(Native Method)

at com.vontu.cracker.jni.NativeContentExtractionEngine.<init>(NativeContentExtractionEngine.java:31)

at com.vontu.cracker.jni.EngineContext.<init>(EngineContext.java:13)

at com.vontu.cracker.NativeExtractionEngine.<init>(NativeExtractionEngine.java:39)

at com.vontu.cracker.NativeExtractionEngine.<init>(NativeExtractionEngine.java:23)

at com.vontu.cracker.NativeExtractionEngine.<init>(NativeExtractionEngine.java:17)

at com.vontu.machinelearning.training.MLDTrainingOrchestrator.initializeExtractionServices(MLDTrainingOrchestrator.java:131)

at com.vontu.machinelearning.training.MLDTrainingOrchestrator.initializeDependencies(MLDTrainingOrchestrator.java:111)

at com.vontu.machinelearning.training.MLDTrainingOrchestrator.orchastrateActions(MLDTrainingOrchestrator.java:61)

at com.vontu.machinelearning.training.MLDTrainingProcess.main(MLDTrainingProcess.java:39)

Cause

The training error may occur when there is insufficient data available to create a valid VML profile.

Resolution

You will need to increase the number of files provided in the negative or positive sets to make a successful VML training profile. The recommended number of documents is 250 per training set. The minimum number of documents per training set is 50

Change the value of logging value from INFO to FINEST in the MLDTrainingLogging.properties file.

For example: com.vontu.machinelearning.training.MLDTrainingLogHandler.level = FINEST

NOTE: For troubleshooting such issues you can increase the logging level in the MLDTrainingLogging.properties file found in the \Program Files\Symantec\DataLossPrevention\EnforceServer\<version>\Protect\config directory on the Enforce server and review the machinelearning_training_operational_x.log where x is the log number found in the \ProgramData\Symantec\DataLossPrevention\EnforceServer\<version>\logs directory:

Once troubleshooting is complete, change the logging value from FINEST back to INFO.