[OPENNLP-155] unreliable training set accuracy in perceptron

• Type: Improvement
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: maxent-3.0.1-incubating
• Fix Version/s: tools-1.5.1-incubating, maxent-3.0.1-incubating
• Component/s: Machine Learning
• Assignee: Jason Baldridge
• Reporter: Jason Baldridge
• Due: 14/Apr/11
• Created: 14/Apr/11 03:24
• Updated: 08/Jun/11 08:01
• Resolved: 08/Jun/11 04:42

Description

The training accuracies reported during perceptron training were much higher than final training accuracy, which turned out to be an artifact of the way training examples were ordered.