Nyby, Clara

Quantifying the performance of machine learning models in materials discovery

In this work, we simulate a sequential learning (SL)-guided materials discovery process and demonstrate a decoupling between traditional model error metrics and model performance in guiding materials discoveries.