Accurate AI prediction even with a small number of experiments

Dmitry Spodarets
Dmitry Spodarets

NIMS, Asahi Kasei, Mitsubishi Chemical, Mitsui Chemicals and Sumitomo Chemical report to have used the chemical materials open platform framework to develop a new AI technique that increases the accuracy of ML-based predictions of material properties by using material structural data from only a small number of experiments. This technique is likely to expedite the development of various materials, including polymers.

The AI technique is capable of selecting potentially promising material candidates for fabrication and accurately predicting their physical properties using XRD, DSC and other measurement data obtained from only a small number of actually synthesized materials.

The use of this technique may enable a more thorough understanding of the relationship between materials' structures and physical properties, which would facilitate investigation of fundamental causes of material properties and the formulation of more efficient materials development guidelines. Furthermore, it is expected to be applicable to the development of a wide range of materials in addition to polyolefins and other polymers, thereby promoting digital transformation (DX) in materials development.