(375c) Can Transfer Learning Assist in Resolving Missing Data for Enhanced Data-Driven Optimization ?

Developing data-driven models has found successful applications in varying engineering problems. In capacitive devices like deionization and supercapacitors, there exists a potential to apply this data-driven machine learning (ML) model in optimizing it potential use in energy efficient separations or generation. However, these models are faced with limited datasets and even in large quantity, the datasets are incomplete, limiting their potential use for successful data-driven modeling. Here, the success of transfer learning in resolving the challenges with limited dataset was exploited. A two-step data-driven ML modeling framework named ImputeNet involving training with ML-imputed datasets and then with clean dataset was explored. Through data imputation and transfer learning, it is possible to develop data-driven model with acceptable metrics mirroring experimental measurement. Using the trained model, genetic algorithm-based optimization study was implemented to analyze the solution in the pareto optimal. This early insight can be used at an initial stage of experimental measurements to rapidly identify experimental conditions worthy of further investigation.