Papayas can be male, female, or hermaphrodite. Usually, papaya growers will select hermaphrodites because they can self-fertilize and thus produce fruit by themselves. However, this selection takes time an resources because the plants have to flower before the sex can determined.

The linked paper is a paper about combining machine learning and spectroscopy in order to determine the sex of a papaya by performing spectral measurements on the seed or leaves.

The spectral differences of the seeds suggest that hermaphrodite seeds have a slightly different lipid composition, but other than that there are no obvious spectral differences between different genders.

The trained machine learning algorithm is also able to discriminate between sexes by measuring the spectra of leaves, even though a careful analysis of the spectra reveals no obvious differences.

As a spectroscopist I would like to have a clear spectral signature that provides some fundamental understanding of the biology of plant sex types. But machine learning algorithms don’t care about these superfluous details, as they can find significant patterns where we might see noise and chaos. It is too bad they can’t communicate to us their insight!