Skip to main content

Table 6 Running time of each classifier

From: LightCpG: a multi-view CpG sites detection on single-cell whole genome sequence data

  Dataset FCNN LightGBM XGBoost RF GBDT
RunningTime (s) HCCs 4138.88 8.30 5.25 2093.28 165.79
  HepG2 1889.55 5.06 2.88 450.05 57.59
  1. 1FCNN (Fully Connected Neural Network): the number of layers was 2, the number of nodes was the number of features, the activation function was sigmoid, the loss function was mean square error and the optimizer was RMSProp
  2. 2LightGBM: the number of trees was 110, the number of max depth was 5, the number of leaves was 22, learning rate was 0.04 and the number of thread was 8. Other parameters were at default values
  3. 3XGBoost: the number of trees was 110, the number of max depth was 7 and the number of thread was 8. Other parameters were at default values
  4. 4RF: the number of trees was 500 and the number of thread was 8. Other parameters were at default values
  5. 5GBDT: the number of trees was 110, the number of max depth was 9 and the number of thread was 8