TRBoost: A Generic Gradient Boosting Machine based on Trust-region Method
share
A generic Gradient Boosting Machine called Trust-region Boosting (TRBoost) is presented for performing supervised machine learning tasks. Existing Gradient Boosting Machines (GBMs) have achieved state-of-the-art results on many problems. However, there are some difficulties to maintain a balance between performance and generality. The first-order algorithms are appropriate for more general loss functions than the second-order algorithms; while the performance is often not as good as the latter one. TRBoost generalizes GBMs based on the Trust-region algorithm to suit arbitrary loss functions while keeping up the good performance as the second-order algorithms. Several numerical experiments are conducted to confirm that TRBoost can get competitive results while offering additional benefits in convergence.
READ FULL TEXT
