Cost-imbalanced hyper parameter learning framework for quality classification

摘要: Abstract A quality control system is an indispensable section in various manufacturing and service industries. It plays a critical role reducing process flaws, optimizing parameters, improving production productivity, as well enhancing customer satisfaction. In this paper, we propose intelligent data-driven classification platform by leveraging novel integrated hyper learning framework to further strengthen the cost-effectiveness economic loss due misclassification. The misclassification-dependent weights are proposed used for training classifier with emphasis on cost-effectiveness. optimally identify such weights. Specifically, consists of two nested layers, where inner-layer addresses optimal given set misclassification weights, while out-layer updates iteratively according performance terms identified towards optimality. case studies implemented using five different datasets industries, including food, auto, steel, glass. loss, additional carbon emission when through framework, compared three other algorithms under settings penalty costs results illustrate that outperforms ones reduction demonstrate robustness respect costs. As reduction, model can outperform, most cases, algorithms. While consistency superiority cannot be guaranteed since environmental concern not modeled objective function.