The intermediate mass Herbig Ae/Be stars are young stars approaching the Main Sequence and are key to understanding the differences in formation mechanisms between magnetic low mass stars and the non-magnetic high mass stars. However; the study of the general properties of these objects is arduous as only around 270 of them are known; with many presenting a doubtful nature.Gaia Data Release 2 has improved and greatly increased the number of sources with an astrometric solution available. In combination with other catalogues; it constitutes a splendid big data breeding ground for applying Machine Learning techniques and algorithms. We present our plan to discover new Herbig Ae/Be stars by first creating a robust training set from this very reduced set of known objects. Several features were chosen for identifying new objects of the class based on our current knowledge of this group; which normally spotlights in infrared excesses; photometric variabilities and Halpha emission lines. This feature selection was complemented with Principal Component Analysis.The training set and the final set of features were used to train a Neural Network; which we later used for looking for new Herbig Ae/Be stars and Pre-Main Sequence objects in general among Gaia DR2 sources. Evaluation on test set concludes that we reach a precision over 90% and a recall over 70%; this meaning that we retrieve more than 7000 new Pre-Main Sequence objects spread all over the galactic plane and; as a side effect; over 1000 classical Be stars; with which they share many characteristics. This in turn will allow us to study the Pre-Main Sequence evolution as a function of mass; age and location in the galaxy to an unprecedented precision.