Skip to Main Content
Table 7 

Comparing different ensemble classifiers against our baselines and oracles. The CV column lists cross-validation within the Toefl11-TrainDev data and the Test column is the Toefl11-Test set. Best ensemble result per column represented in bold.

 MethodAccuracy (%)
CVTest
Baselines Random Baseline 9.1 9.1 
Single Vector Baseline 78.2 77.5 
  
2013 Shared Task Winner 84.5 83.0 
Bykh and Meurers (2014) — 84.8 
Ionescu, Popescu, and Cahill (2014) 84.1 85.3 
  
Oracles Accuracy@2 91.8 92.0 
Accuracy@3 94.5 94.6 
Oracle 96.1 96.0 
  
Ensembles Plurality Voting 82.6 82.5 
Borda Count 81.2 81.5 
Mean Probability 82.6 83.3 
Median Probability 82.4 82.7 
Product Rule 80.3 80.6 
Highest Confidence 80.1 80.4 
 MethodAccuracy (%)
CVTest
Baselines Random Baseline 9.1 9.1 
Single Vector Baseline 78.2 77.5 
  
2013 Shared Task Winner 84.5 83.0 
Bykh and Meurers (2014) — 84.8 
Ionescu, Popescu, and Cahill (2014) 84.1 85.3 
  
Oracles Accuracy@2 91.8 92.0 
Accuracy@3 94.5 94.6 
Oracle 96.1 96.0 
  
Ensembles Plurality Voting 82.6 82.5 
Borda Count 81.2 81.5 
Mean Probability 82.6 83.3 
Median Probability 82.4 82.7 
Product Rule 80.3 80.6 
Highest Confidence 80.1 80.4 
Close Modal

or Create an Account

Close Modal
Close Modal