The classical gradient descent method is generalized without using subgradient information of objective ... Rates of convergence are comprehensively explored for the algorithm with RWA (DRGFMD-RWA).

Abstract: In this article, we propose a novel entropy and confidence-based undersampling boosting (ECUBoost ... algorithms without iterations such as decision trees. Meanwhile, random forests are used ...

This repository contains implementations of the Random Cut Forest (RCF) probabilistic data structure. RCFs were originally developed at Amazon to use in a nonparametric anomaly detection algorithm for ...