|Title||RLT: Reinforcement learning trees (R).|
|Year of Publication||2015|
|Authors||Zhu, Ruoqing, and Michael R. Kosorok|
RLT Implements reinforcement learning trees with corresponding features, and a variety of existing tree-based methods such as random forests and extremely randomized trees. The parameter reinforcement enables reinforcement learning which will search a global best variable as the splitting variable at each internal node. This is achieve by fitting an embedded forest model to calculate and compare the variable importance measure. Shared memory parallel computing through "openMP" is enabled by specifying the use_cores parameter to > 1. A variety of other features are also implemented such as subject weights and variable weights when searching for the splitting variable.
|Original Publication||RLT: Reinforcement learning trees (R).|