Daniel El Ouraoui, Pascal Fontaine and Cezary Kaliszyk

We provide here experimental data.

Solver:

The binaries files of veriT here

• The verbose mode can be activated using the following option --verb-learning-stats
  
  

• Activate veriT without learning in comparative mode:
  
  

  ./veriT --disable-print-success --compare-ml-native

• Activate the learning option in veriT with the first model veriT(M):
  
  

  ./veriT --disable-print-success --learning-native

• Activate the learning option in veriT with the second model veriT(M²):
  
  

  ./veriT --disable-print-success --learning-native --learning-native-2

• Activate the learning option in veriT with both models veriT(M + M²):
  
  

  ./ml-m2.sh

• Activate veriT learning that is veriT + portofolio:
  
  

  ./ml-without-learning.sh

• Activate the learning option in veriT with a portfolio approach veriT(M + M²) + portofolio:
  
  

  ./ml-with-learning.sh

Benchmarks:

• List of problems used to train veriT(M)
  
  

  Training
  
  

  Test

• List of problems used to train veriT(M²)
  
  

  Training
  
  

  Test
  
  

XGBoost configuration:

• XGBoost training configuration for veriT(M) here
  
  

• XGBoost training configuration for veriT(M²) here
  
  

Results:

The evaluations here

---