Below are several major papers and resources that define the field:
: A few parameter combinations ("stiff") tightly constrain model behavior, while others ("sloppy") can vary by orders of magnitude without changing the output. sloppy
(Waterfall et al., 2006): Proposes that sloppy models belong to a common "universality class" with eigenvalue spectra that are roughly constant on a logarithmic scale. Below are several major papers and resources that
: Researchers use the FIM to measure how distinguishable models are based on their predictions. In sloppy models, FIM eigenvalues are distributed roughly evenly over many decades. sloppy