FLEXIBLE METAMODELS: HARVESTING SIMULATION DATA
Year: 2017- present
Role: Doctoral thesis work
As AI makes its way through to design offices, we also need to implement new ways of generating, organising and harvesting data with each project, such that it can be used for subsequent projects. In a recent study, I developed a preliminary approach to collect and harvest simulation data from one finite element analysis study to the next, such that the data can be used cumulatively to build subsequent statistical or machine learning models. The approach was inspired by the notion of ‘transfer learning’ from neural networks.