B1131
Title: Combining design of experiments and machine learning in industrial experiments
Authors: Roberto Fontana - Politecnico di Torino (Italy)
Luigi Salmaso - University of Padova (Italy)
Alberto Molena - Università Degli studi di Padova (Italy) [presenting]
Abstract: In product innovation, an emerging trend is represented by the combined utilization of Machine Learning (ML) models and Design of Experiments (DOE) techniques. The purpose of the contribution is two-fold: firstly, to assess the most fitting designs and ML models to be used together in an active learning approach, and then to present ALPERC, an iterative approach based on non-parametric ranking and clustering suitable for physical experiments when two or more responses are investigated. The validity of the approach will be tested using simulation studies and a real case study about the construction and refinement of a multi-response emulator to estimate three critical temperatures in some innovative metallic alloys.
