Automated Modeling Of Phytoplankton GrowthUsing Ecological Domain Knowledge

Abstract

Using ecological domain knowledge, machine discovery systems can help human experts to generate models from measured data. In contrast with traditional modeling methods, which are used to identify parameter values of the model with prescribed structure, machine learning tools identify the structure of the model as well. In the paper, we present LAGRAMGE, an equation discovery system that uses context free grammars to define the space of possible model structures, and can also make use of domain specific background knowledge in the form of function definitions. We use LAGRAMGE to automate the modeling of phytoplankton growth in Lake Glumsoe, Denmark. The structure of the automaticly constructed model agrees with human experts expectations. The model can be successfully used for short-term prediction of the phytoplankton concentration.