GPU-Accelerated CatBoost-Forest for Hyperspectral Image Classification Via Parallelized mRMR Ensemble Subspace Feature Selection

In this article, the graphics processing unit (GPU)-accelerated CatBoost (GPU-CatBoost) algorithm for hyperspectral image classification is first introduced and comparatively studied using diverse features. To further foster performance from both accurate efficient viewpoints, an ensemble version of GPU-CatBoost, GPU-accelerated CatBoost-Forest (GPU-CatBF) algorithm, proposed by adopting parallelized minimum redundancy maximum relevance (mRMR) (PmRMRE) feature selection (FS) algorithm. evaluate suitability mRMR PmRMRE, 11 other state-of-the-art FS algorithms are comprehensively investigated. Experimental results on three widely acknowledged benchmarks showed that: 1) GPU-CatBoost also advanced learning (EL) features; 2) PmRMRE have properties highly discriminative features band selection, best achieved in most cases terms robustness computational efficiency; 3) GPU-CatBF always outperforms while compatible even better reachable without losing much efficiency contrast with selected decision tree-based EL algorithms.