Parametric and non-parametric gradient matching for network inference: a comparison.

We found that parametric methods can provide comparable, and often improved inference compared to non-parametric methods; the latter, however, require no kinetic information and are computationally more efficient.