HP5 - Materials Informatics

The goal of HP5 is to develop data driven methodologies for the acceleration of material discovery. The scientific community has been generating enormous amounts of valuable knowledge that has mostly remained confined into papers. Recently, it became evident that the data driven approach can unlock the huge potential of this data and allow for drastic acceleration of discovery. Indeed, HP5 researchers demonstrated that machine learning models can provide very accurate predictions of materials properties at a fraction of the cost and time needed by traditional approaches. Extracting the scientific literature treasure and structuring into searchable forms allow researchers to very fast learn from the collective experience of the community and thus move forward much faster. This is of course only the beginning. We have merely scratched the surface and already gotten strong evidence of the full potential of the approach.