FastBDT A speed optimized and cache friendly implementation of stochastic grad

FastBDTFastBDT A speed-optimized and cache-friendly implementation of stochasticgradient-boosted decision trees for multivariate classificationThomas Keck thomas.keck2@kit.edu Karlsruher Institut für Technologie, Campus Süd Institut für Experimentelle Kernphysik Wolfgang-Gaede-Str. 1 76131 KarlsruheEditors:AbstractStochastic gradient-boosted decision trees are widely employed for multivariate classifi- cation and regression tasks. This paper presents a speed-optimized and cache-friendly implemen