{"id":"arxiv/2611.11914","url":"https://arxiv.org/abs/2611.11914","tags":["cs.LG","math.ST"],"score":0,"title":"CRAFT-GBM: Selecting Best Features for Gradient-Boosted Machine Learning","domain":"cs.LG","source":"arxiv","authors":["Basil Kuok","Yi Zhou","Yao Xuan","Shuping Li","Ke-Wei Huang"],"reviews":0,"subject":"cs.LG","summary":"Feature subsampling is used to accelerate gradient-boosted trees, yet standard approaches fix the feature count in advance and largely ignore the split-gain evidence accumulated during training. We introduce CRAFT-GBM (Confidence-Ranked Adaptive Feature Tracking), a native LightGBM extension that uses this evidence to adapt which---and, by default, how many---features are searched each boosting round. CRAFT-GBM transforms root-split gains into bounded observations and constructs joint time-uniform bounds valid under adaptive, non-stationary training histories. In its default endogenous mode,…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-11-09","reactions":{}}