H. Kaneko
Meiji University,
Japan
Keywords: machine learning, direct inverse analysis, materials design
Summary:
Abstract In molecular, material, and process designs, it is important to perform inverse analysis of mathematical models constructed with machine learning using target values of the properties and activities. While many approaches employ a pseudo-inverse analysis, Gaussian mixture regression (GMR) can achieve direct inverse analysis. Although Bayesian optimization (BO) is an effective tool, BO merely selects a candidate from a limited number of samples, and the samples do not necessarily contain the optimal solution. Furthermore, because upper and lower limits are set for explanatory variables x, it is not possible to obtain solutions that go beyond these limits. To solve these issues, direct inverse analysis of the GMR model was proposed because GMR models can estimate x values directly based on the target values of objective variables y (Chemom. Intell. Lab. Syst. (2021) 104226). The proposed method could allow the target y value to be achieved with a dramatically smaller number of experiments than by BO, especially when the number of x-variables was large. Furthermore, the proposed direct inverse analysis was applied to time-series data analysis and process design, and could design both the batch time and the process variable profiles (x) to ensure that the endpoints (y), such as the product quality and the material properties, possess the target values following a batch process (Comput. Chem. Eng. (2023) 108072). Biography I got my PhD in Chemical System Engineering from the University of Tokyo, Japan in September 2011, including a significant research at Imperial College London, UK, from January to March 2010. In October 2011, I commenced my professional career as an assistant professor in chemical system engineering at the University of Tokyo. In April 2017, I transitioned to Meiji University, Japan, serving as an associate assistant professor in the department of applied chemistry until March 2020. My role extends beyond academia with several concurrent appointments. Since December 2018, I have been a Guest Principal Researcher at RIKEN. I also hold positions as a Visiting Associate Professor at Osaka University and Hiroshima University since April 2019. In the corporate sector, I have been the President of Future Science Research Institute LLC since November 2020 and the Chief Technical Officer (CTO) at Data Chemical Co., Ltd., since October 2021. My contributions to the field are acknowledged by numerous awards, including the Chemical Engineering Society Research Encouragement Award in 2015 and the Symposium Award at the International Congress on Pure & Applied Chemistry in 2023. Additionally, I have engaged in editorial and advisory roles, serving as an editor for the Journal of Chemical Engineering of Japan and as a committee member in various NEDO research projects.