Skip to Main Content
Table 7 
Analysis of refinement methods applied to the same initial mappings of our adversarial autoencoder on the Conneau data set for Es, De, It, and Fi languages.
 En-EsEn-DeEn-ItEn-Fi
Artetxe refinement 
Robust self-learning 82.3 84.0 75.3 73.6 78.7 78.9 46.7 60.8 
Robust self-learning + Symmetric re-weighting 82.7 84.7 75.4 74.1 79.2 79.4 49.4 64.6 
  
Our refinement 
Procrustes solution 82.6 84.5 75.5 73.9 78.8 78.9 49.1 64.0 
Symmetric re-weighting 82.8 84.8 75.7 74.5 79.1 80.0 47.6 64.0 
Procrustes solution + Symmetric re-weighting 83.0 85.2 76.2 74.7 79.3 80.3 49.8 65.7 
 En-EsEn-DeEn-ItEn-Fi
Artetxe refinement 
Robust self-learning 82.3 84.0 75.3 73.6 78.7 78.9 46.7 60.8 
Robust self-learning + Symmetric re-weighting 82.7 84.7 75.4 74.1 79.2 79.4 49.4 64.6 
  
Our refinement 
Procrustes solution 82.6 84.5 75.5 73.9 78.8 78.9 49.1 64.0 
Symmetric re-weighting 82.8 84.8 75.7 74.5 79.1 80.0 47.6 64.0 
Procrustes solution + Symmetric re-weighting 83.0 85.2 76.2 74.7 79.3 80.3 49.8 65.7 
Close Modal

or Create an Account

Close Modal
Close Modal