: this is probably a quite serious piece of work, and it's been published in a refereed journal. But I can't help but notice their, er, taste in choosing the subject matter of their work.
And that warning has broken, I guess, the punchline. What I wanted to do is to link to the paper Pork farm odour modelling using multiple-component multiple-factor analysis and neural networks
I would say a single neuron is needed: one that says "It f*cking stinks to high heaven" every little while. But I guess that's not too scientific. Future line of works might include: "Sweaty armpit smell analysis using Radial Basis Functions" and "Belly button lint deconstruction via expert systems".