Landmark-Based Ensemble Learning with Random Fourier Features and Gradient Boosting

This paper jointly leverages two state-of-the-art learning stra-tegies—gradient boosting (GB) and kernel Random Fourier Features (RFF)—to address the problem of learning. Our study builds on a recent result showing that one can learn distribution over RFF to produce new suited for task at hand. For this distribution, we exploit GB scheme expressed as ensembles weak learners, each them being function designed fit residual. Unlike Multiple Kernel Learning techniques make use pre-computed dictionary functions select from, iteration by approximating it from training data weighted sum RFF. strategy allows build classifier based small ensemble learned “landmarks” better underlying application. We conduct thorough experimental analysis highlight advantages our method compared both boosting-based kernel-learning methods.