Abstract:The inequality constraint of stochastic configuration networks (SCNs) is the key to guarantee its universal approximation capability. SCNs add new units to neural networks incrementally under its supervisory mechanism. However, with the increasing number of hidden nodes, the output weights which are evaluated by the least squares method are frequently unstable. This unstable phenomenon (arbitrarily small perturbations may lead to large changes in the solutions) can be classified as ill-posed problems. One of the most important properties of ill-posed problems is instability and the widely used Tikhonov regularization method to deal with ill-posed problems is undesirable. This paper studies the supervisory mechanism of SCNs and algebraic properties of the hidden output matrix. Then a new greedy stochastic configuration network (GSCN) for ill-posed problems is presented. Firstly, under the premise of ensuring the universal approximation capability, GSCN tries to optimize the randomly assigned input weights and biases based on hunter–prey optimization (HPO) algorithm. The optimized hidden parameters which can reduce the residual error most are selected for the newly added hidden node. Secondly, the singular value decomposition (SVD) and orthogonal-triangular (QR) decomposition with column pivoting are introduced to extract linearly independent subset of the hidden output matrix for ill-posed problems of SCNs. Finally, one function approximation, six benchmark regression, two classification datasets and bearing fault diagnosis are employed to evaluate the performance of GSCN. Experimental results indicate that GSCN shows superior performance with respect to fast convergence, generalization and stability power than other contrast methods. The code is available at https://github.com/Gzu-public-bigdata/GSCN .

Multitarget Robust Deep Stochastic Configuration Network Parameter Modeling Method

Robust Multi-Target Regression with Improved Stochastic Configuration Networks and Its Applications

Stochastic Configuration Network Modeling Method Based on Information Superposition and Mixture Correntropy

Multi-target regression via stochastic configuration networks with modular stacked structure

Robust Stochastic Configuration Networks with Maximum Correntropy Criterion for Uncertain Data Regression.

Enhancing Deep Stochastic Configuration Networks: Efficient Training via Low-Rank Matrix Approximation

Deep Stochastic Configuration Networks with Optimised Model and Hyper-Parameters

Multi-Target Regression Via Robust Low-Rank Learning.

Robust Stochastic Configuration Networks with Kernel Density Estimation for Uncertain Data Regression

An AdaBoost Based - Deep Stochastic Configuration Network.

Adaptive inertia weights: an effective way to improve parameter estimation of hidden layer in stochastic configuration networks

Distributed Learning of Predictive Structures from Multiple Tasks over Networks

Deep Stochastic Configuration Networks with Different Random Sampling Strategies

Deep Stochastic Configuration Networks with Universal Approximation Property

Efficiency optimization methods for stochastic configuration networks

The Balanced Multi-Modal Spiking Neural Networks with Online Loss Adjustment and Time Alignment

Dynamic Optimization of Neural Network Structures Using Probabilistic Modeling

Deeper Insights into Learning Performance of Stochastic Configuration Networks

Greedy stochastic configuration networks for ill-posed problems

Proximal Multitask Learning Over Distributed Networks With Jointly Sparse Structure

A regularized stochastic configuration network based on weighted mean of vectors for regression