Machine Learning Aided Inverse Design for New Wine

Wine quality for physicochemical properties is a high dimensional complex network, which how to accurately understand the internal relationship and design to meet requirements new wine is a challenging problem. In this paper the causality of correlation between the physicochemical properties is detected by causal discovery (PCMCI), and the features importance interpret magnitude based on interpretability machine learning model. Inverse design methods can start from a desired property and optimizing a corresponding internal structure. Here, we combine machine learning model and reverse design have proposed to for both the forward and inverse problem for physicochemical properties and wine quality. Our work can accelerate the design process and reduce cost. According to our knowledge, this study is the first to offer easily understandable design guidelines for quickly locating approximations that can be used in the creation of new wines.