The excess of gamma rays detected by the Fermi-LAT telescope, first observed more than 15 years ago at the centre of our Galaxy, remains one of the most fascinating mysteries in astrophysics. Its origin remains unknown and continues to fuel much debate: could it be the first signal from dark matter, or is it the sign of an as yet unknown population of gamma-ray emitters? Millisecond pulsars (MSPs), rapidly rotating neutron stars located in the Galactic bulge, are among the most promising candidates for explaining this excess. These old objects are mainly identified by deep radio surveys, but understanding their distribution and emission mechanisms is a complex challenge. The objective of this M2 project is to apply advanced statistical techniques to the wealth of data available, so to reveal the underlying properties of MSP populations, in particular their spatial and luminosity distribution in the Galaxy. This is where machine learning comes in, offering a revolutionary approach to analysing and interpreting this abundance of data.