I’m really happy to have presented our recent work with Nikos Dervilis and Keith Worden, “On the use of Symbolic Regression for Population-Based Modelling of Structures”.

In this work, we attempt to tackle the problem of knowledge transfer for regression in a one-to-one scenario, i.e. to transfer knowledge from an extensively–monitored structure to a new testing structure. We use experimental data from aluminium plates and we show how with our method we were able to perform the transfer of the regression model. There is still more work to be done, especially if we want to include some physical knowledge in the inference procedure and to take uncertainty into account.

You can read the paper here or have a look at the slides of the presentation here.