Unraveling the Interconnected Axes of Heterogeneity in Machine Learning for Democratic and Inclusive Advancements