Ensemble Deep Random Vector Functional Link Neural Network for Regression

Inspired by the ensemble strategy of machine learning, deep random vector functional link (dRVFL), and dRVFL (edRVFL) has shown state-of-the-art results on different datasets. Our present work first fills gap edRVFL in field regression. We test evaluate performances dRVFLs regression problems. Subsequently, we propose a novel regularization method boosted factor (BF), two variants with skip connection (edRVFL-SC) connections (edRVFL-RSC) one (esc-edRVFL) which show significant improvement over original dRVFL. The BF is newly introduced hyperparameter to scale values activated hidden neurons accommodate diversity data, it also able filter neurons. edRVFL-SC edRVFL-RSC are connections. In edRVFL-SC, apply dense edRVFL, inspired residual architecture learning area. However, due specificity randomized networks, simple probably leading reuse useless features. To address this problem, connection-based can keep latent space. esc-RVFL an scheme that utilizes several models trained folds training dataset. esc-edRVFL identified as best-performing algorithm through comprehensive evaluation 31 UCI