AI 용어집
인공지능 완전 사전
Random Forest
Bagging algorithm using decision trees as base estimators, where each tree is trained on a bootstrap sample and a random subset of features.
Out-of-Bag (OOB) Error
Method for evaluating the performance of a Bagging model by using the data not selected during bootstrap sampling for each estimator as a validation set.
Estimator Instability
Property of certain base models (such as decision trees) that are very sensitive to small variations in data, making Bagging particularly effective for them.
Random Patches
Ensemble technique combining bootstrap sampling on instances and random sampling on features for each base model, increasing diversity.
Bagging (Regression)
Specific application of Bagging to regression models, where final predictions are obtained by calculating the average of individual predictions from base regressors.
Ensemble Diversity
Measure of the heterogeneity of errors between base models, with high diversity being crucial for aggregation in Bagging to effectively reduce overall error.