Bigdata/AI Solutions for Materials Development
-
Researchers:
Kwang-Ryeol Lee, Sang Soo Han, Jung-Hoon Lee, Donghun Kim, Byungju Lee
-
Representative Achievement:
• KiRI (KIST Research Informatics) System to enable bigdata-driven research environments
• Technology Transfer of an ML model to Virtual Lab, Inc. (ML model to predict adsorption energy in catalysis)
Mission Description
Recently, as the successful stories for development of novel materials using computer simulations has increased, an interest in computer simulations is also increasing. The material design using computer simulations is a method to predict the physical/chemical properties from material information (composition, structure, etc.). Thus, it can be called a principle-based research method based on “causality” such as quantum/classical/statistical mechanics.
Material researchers have accumulated material data through countless experiments and theoretical studies so far. The method, finding the “correlation” between information of materials and their properties using material database (DB) and artificial intelligence (AI) and then designing novel materials based on the correlation, has been actively studied in recent years. Here, no principles such as quantum/classical/statistical mechanics are utilized. Due to this, the big data/AI-driven technology for material design can provide a new paradigm in material development.
Computational Science Research Center (CSRC) aims to lead the field of data-driven material design by systematically building DBs of materials and developing various AI technologies for materials. While computer simulations basically predicts material properties from material information, big data/AI technologies can realize the “inverse” design which is a method to design novel materials from targeted material properties. Moreover, big data/AI technologies can be utilized to understand the correlations between materials, properties and processes. CSRC is developing the inverse design technologies and the synthesis planning technologies that predict the synthesis or process conditions of a target material. Development of AI technologies for materials is impossible without material DBs. CSRC is also focusing on development of natural language process technologies that efficiently and accurately extracts material data from literatures and then building material DBs. The DB and AI technologies will realize the 4th industrial revolution-typed future laboratories for materials development though linkage with the Autonomous Labs.
