Techniques for relating complex vibrational observables with physical variables that are readily available in molecular simulations have been used extensively in the past two decades to connect to vibrational spectroscopy experiments with great success. However, the simplicity of maps based mainly on linear regression of electrostatic factors limits their maximum accuracy. Starting with vibrational modes for which successful maps have not been created, we will endeavor to deeply understand the character of the vibrations using quantum chemistry calculations. We will then combine these with molecular dynamics simulations, producing the large amounts of data required to build an accurate map. Finally, using advanced machine learning techniques, we build new maps and apply them to uncover powerful new insights about complex systems. In this aim, a current collaborative project involves work with Lou Charkoudian and Casey Londergan at Haverford to build a map for the Raman active alkyne stretching mode, which will be used to understand the dynamics of carrier proteins.