This thesis explores the fast evaluation of supersymmetric cross sections using Gaussian processes — a machine learning method for regression. The time- consuming nature of accurate evaluation of higher order cross sections has been a limiting factor in searches for supersymmetry at the LHC and elsewhere. The recent proposition of distributing Gaussian processes between individual estima- tors, in the form of a robust Bayesian Comittee Machine, allows for the use of larger datasets than before, thus putting the desired accuracy for regression within reach. The distributed Gaussian processes also allow for parallelisation to make evaluation even faster. A distributed Gaussian model for squark pair production in proton-proton collisions was built and tested, predicting cross sections within 10% of next-to-leading order cross sections calculated by Prospino 2.1 in a fraction of the time.