Abstract:Stochastic Configuration Networks (SCNs) are a class of randomized neural networks that integrate randomized algorithms within an incremental learning framework. A defining feature of SCNs is the supervisory mechanism, which adaptively adjusts the distribution to generate effective random basis functions, thereby enabling error-free learning. In this paper, we present a comprehensive analysis of the impact of the supervisory mechanism on the learning performance of SCNs. Our findings reveal that the current SCN framework evaluates the effectiveness of each random basis function in reducing residual errors using a lower bound on its error reduction potential, which constrains SCNs' overall learning efficiency. Specifically, SCNs may fail to consistently select the most effective random candidate as the new basis function during each training iteration. To overcome this problem, we propose a novel method for evaluating the hidden layer's output matrix, supported by a new supervisory mechanism that accurately assesses the error reduction potential of random basis functions without requiring the computation of the Moore-Penrose inverse of the output matrix. This approach enhances the selection of basis functions, reducing computational complexity and improving the overall scalability and learning capabilities of SCNs. We introduce a Recursive Moore-Penrose Inverse-SCN (RMPI-SCN) training scheme based on the new supervisory mechanism and demonstrate its effectiveness through simulations over some benchmark datasets. Experiments show that RMPI-SCN outperforms the conventional SCN in terms of learning capability, underscoring its potential to advance the SCN framework for large-scale data modeling applications.

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

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

Multi-Target Regression Via Robust Low-Rank Learning.

Multi-target Regression Via Target Specific Features

Learning local instance correlations for multi-target regression

An Ensemble-Adaptive Tree-Based Chain Framework for Multi-Target Regression Problems

Multitarget Robust Deep Stochastic Configuration Network Parameter Modeling Method

Performing Multi-Target Regression via a Parameter Sharing-Based Deep Network

Multi-target Regression Via Self-Parameterized Lasso and Refactored Target Space

Broad Stochastic Configuration Network for Regression

A Stochastic Configuration Network with Attenuation Regularization and Multi-kernel Learning and Its Application

A holistic global-local stochastic configuration network modeling framework with antinoise awareness for efficient semi-supervised regression

Sparse Multi-Relational Graph Convolutional Network for Multi-type Object Trajectory Prediction

Recurrent Stochastic Configuration Networks with Hybrid Regularization for Nonlinear Dynamics Modelling

Learning Structured Components: Towards Modular and Interpretable Multivariate Time Series Forecasting

Deeper Insights into Learning Performance of Stochastic Configuration Networks

Distributed Stochastic Configuration Networks with Cooperative Learning Paradigm

Kernel Stochastic Configuration Networks for Nonlinear Regression

Multi-attention Associate Prediction Network for Visual Tracking

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

Stochastic Configuration Networks: Fundamentals and Algorithms