Molecules, Materials, and Products

The properties of formulated and functionalized products depend heavily on the operating conditions prevailing during their synthesis. Modeling the relationship between process and properties is therefore crucial for ensuring product quality and optimizing production. Depending on the characteristics and complexity of the process, phenomenological models may be sufficiently accurate to reduce the need for data-driven models. The state of the art nevertheless highlights numerous applications of machine learning methods for predicting product properties based on processing conditions. The inverse procedure—that is, predicting processing conditions based on target properties—is also encountered, though less frequently.

One industrial sector with a wide range of applications is that of polymers: the relationship between process and properties is a major concern in this...