Development of efficient algorithms for Single-pass neural networks

Development of efficient algorithms for Single-pass neural networks

Automatic regularization parameter selection using MRM for Regularized ELM

GCNN*

Automatic smoothing parameter selection using BAT algorithm in Laplacian Weighted PNN for Classification

regularization parameter

Number of hidden units

L1-ELM and L2-ELM

L1-ELM

Novel weighting approach

Laplace distribution

Smoothing factor estimation using BAT algorithm

Laplace distribution

Clustering approach

Estimation of smoothing parameter

Automatic regularization parameter selection for L1-norm regularized ELM using L-curve and U-curve

IRLS

Automatic estimation of regularization parameter using L-curve and U-curve method for L2-norm regularized ELM

L-curve

U-curve

Compared with the GCV and Pseudo-inverse and CV

ADMM

ISTA and FISTA

click to edit