Decomposing functions in terms of higher-order harmonics is a central topic in higher-order Fourier analysis. In its simplest form, such a decomposition is as follows. For a bounded function defined on a finite abelian group , we write it as where: is the sum of « a few » Fourier characters with large amplitudes, is a function whose largest Fourier amplitude is « small » (which is the same as having a small Gowers norm), and is small in . Higher-order analogues where we ask to be small in the Gowers norm for are interesting as we may use them to, e.g., prove Szemerédi’s theorem with good quantitative bounds. Many results guarantee that such a decomposition exists, but few are implementable in applied scenarios. In this talk, we will present a practical approach to finding such a decomposition in the case and demonstrate its performance on synthetic data.