DART: Dropout Regulation in Boosting Ensembles

The dropout approach developed by Hinton has been widely employed in the context of deep learnings to prevent the deep neural network from over-fitting, as shown in https://statcompute.wordpress.com/2017/01/02/dropout-regularization-in-deep-neural-networks.

In the paper http://proceedings.mlr.press/v38/korlakaivinayak15.pdf, the dropout is also proposed to address the over-fitting in tree boosting ensembles, e.g. MART, caused by the so-called “over-specialization”. In particular, while first few trees added at the beginning of ensembles would dominate the model performance, the rest trees added later can only contribute to learning from a small subset, which increases the risk of over-fittings. The idea of DART is to build an ensemble by randomly dropping boosting tree members. The percentage of dropouts would determine the degree of regularization for tree ensembles.

Below is a demonstration showing the implementation of DART in the R xgboost package. First of all, after importing the data, we divided it into two pieces, one for training and the other for testing.

pkgs <- c('pROC', 'xgboost')
lapply(pkgs, require, character.only = T)
df1 <- read.csv("Downloads/credit_count.txt")
df2 <- df1[df1$CARDHLDR == 1, ]
set.seed(2017)
n <- nrow(df2)
sample <- sample(seq(n), size = n / 2, replace = FALSE)
train <- df2[sample, -1]
test <- df2[-sample, -1]

For the comparison purpose, we first developed a tree boosting ensemble without dropouts, as shown below. For the simplicity, all parameters were chosen heuristically. The max_depth is set to 3 due to the fact that the boosting tends to work well with weak learners. While the ROC for the training set can be as high as 0.95, the ROC for the testing set is only 0.60 in our case, implying the over-fitting issue.

mart.parm <- list(booster = "gbtree", nthread = 4, eta = 0.1, max_depth = 3, subsample = 1, eval_metric = "auc")
mart <- xgboost(data = as.matrix(train[, -1]), label = train[, 1], params = mart.parm, nrounds = 500, verbose = 0, seed = 2017)
pred1 <- predict(mart, as.matrix(train[, -1]))
pred2 <- predict(mart, as.matrix(test[, -1]))
roc(as.factor(train$DEFAULT), pred1)
# Area under the curve: 0.9459
roc(as.factor(test$DEFAULT), pred2)
# Area under the curve: 0.6046

With the same set of parameters, we refitted the ensemble with dropouts, e.g. DART. As shown below, by dropping 10% tree members, the ROC for the testing set can be improved from 0.60 to 0.65. In addition, the performance disparity between training and testing sets with DART decreases significantly.

dart.parm <- list(booster = "dart", rate_drop = 0.1, nthread = 4, eta = 0.1, max_depth = 3, subsample = 1, eval_metric = "auc")
dart <- xgboost(data = as.matrix(train[, -1]), label = train[, 1], params = dart.parm, nrounds = 500, verbose = 0, seed = 2017)
pred1 <- predict(dart, as.matrix(train[, -1]))
pred2 <- predict(dart, as.matrix(test[, -1]))
roc(as.factor(train$DEFAULT), pred1)
# Area under the curve: 0.7734
roc(as.factor(test$DEFAULT), pred2)
# Area under the curve: 0.6517

Besides rate_drop = 0.1, a wide range of dropout rates have also been tested. In most cases, DART outperforms its counterpart without the dropout regulation.