Skip to main content

Table 7 Error rates (top1, top3) for the fold recognition problem.

From: Building multiclass classifiers for remote homology detection and fold recognition

   fd25 fd40
   top1 top3 top1 top3
Two-Level Approaches Without Hierarchy Information      
  Scaling 38.5 24.5 25.6 15.4
SVM-Struct Scale & Shift 37.4 24.8 24.7 15.1
  Crammer Singer 36.3 22.7 25.0 13.4
Two-Level Approaches With Class-level Nodes      
  Scaling 36.7 21.9 20.6 11.9
SVM-Struct Scale & Shift 36.3 21.6 21.2 12.2
  Crammer Singer 37.1 22.3 25.3 13.4
Two-Level Approaches With Superfamily-level Nodes      
  Scaling 39.9 24.5 27.9 19.5
SVM-Struct Scale & Shift 39.6 23.4 25.3 16.0
  Crammer Singer 40.6 27.3 26.7 15.1
Two-Level Approaches With Superfamily-level and Class-level Nodes      
  Scaling 39.2 25.2 20.6 13.7
SVM-Struct Scale & Shift 38.5 23.0 20.9 12.2
  Crammer Singer 37.1 23.7 24.1 12.5
  1. The results shown in the table are optimized for the balanced loss function.