FLAIR-THU / VFLAIR

THU-AIR Vertical Federated Learning general, extensible and light-weight framework
77 stars 17 forks source link

Implementation of T-VFL #1

Closed Koukyosyumei closed 1 year ago

Koukyosyumei commented 1 year ago

Implement XGBoost and RandomForest for T-VFL.

benchmark on breast_cancer (main_tree.py)

type of model: xgboost, encryption:False
 training time: 0.15423321723937988 [s]
 train auc: 0.9830216247808299, test auc: 0.9882200567410879
type of model: xgboost, encryption:True
 training time: 0.6558055877685547 [s]
 train auc: 0.9830216247808299, test auc: 0.9882200567410879
type of model: randomforest, encryption:False
 training time: 0.46831250190734863 [s]
 train auc: 0.9630333138515489, test auc: 0.930615517454052
type of model: randomforest, encryption:True
 training time: 5.1531431674957275 [s]
 train auc: 0.9630333138515489, test auc: 0.930615517454052