Modelling paradigm

machine-learning
Modelling paradigm

In the modelling framework, normally the space of all function $f: A \rightarrow B \in Func(A,B)$ is much too large. Therefore we choose some modelling paradigm $M \subset Func(A,B)$ to pick our candidates from. For example, in linear regression, we choose to pick linear functions is some predefined set of inputs.