Real-time predictive modeling is highly anticipated in the field of industrial artificial intelligence (IAI), where neural networks play a crucial role. To effectively model and save data size for industrial applications, this paper introduces a new randomized learner model called stochastic configuration machines (SCMs), which is based on stochastic configuration networks (SCNs). Unlike SCNs and random vector functional-link (RVFL) nets with binarized implementation, the model storage of SCMs can be significantly compressed while still maintaining favorable prediction performance. Additionally, this contribution includes the architecture of the SCM learner model, its learning algorithm, and a theoretical analysis of the learning capacity of SCMs. Experimental studies conducted on benchmark datasets and three industrial applications demonstrate the potential of SCM for industrial data analytics.
Stochastic Configuration Machines: Enabling Industrial Artificial Intelligence (arXiv:2308.13570v1 [cs.LG])
