Machine learning meets network science: dimensionality reduction for fast and efficient embedding of networks in the hyperbolic space