POLAR is a scientific research project that intends to develop the next generation of mathematical signal processing, machine learning and data analysis methods for the Big Astrophysical Data era. So far, the existing techniques used by astronomers have not been sufficient to resolve the problem of identifying intrinsically polarized stars, and the subject of intrinsic stellar polarizations is vastly underexplored. On the other hand, in the era of large polarization datasets, the need for alternative and automated tools for this task is inevitable. The innovative paradigm of POLAR will capitalize on the power and efficiency of cross-modal machine learning (ML) techniques and deep learning (DL) architectures for solving highly complex inverse and large-scale optimization problems. POLAR’s computational intelligence platform will be tested on already existing polarization measurements performed by the RoboPol instrument on the 1.3m telescope of the Skinakas Observatory combined with the stellar distances provided by ESA’s Gaia mission, and will be upscaled to later analyze the future, much larger PASIPHAE polarization datasets.