Abstract Third-generation ground-based gravitational wave detectors are expected toobserve O(10 5 ) of overlapping signals per year from a multitude of astrophysicalsources that will be computationally challenging to resolve individually. Onthe other hand, the stochastic background resulting from the entire populationof sources encodes information about the underlying population, allowing forpopulation parameter inference independent and complementary to that obtainedwith individually resolved events. Parameter estimation in this case is stillcomputationally challenging, as computing the power spectrum involves sampling∼ 10 5 sources for each set of hyperparameters describing the binary population.In this work, we build on recently developed importance sampling techniquesto compute the SGWB efficiently and train neural networks to interpolate theresulting background. We show that a multi-layer perceptron can encode themodel information, allowing for significantly faster inference. We test the networkassuming an observing setup with CE and ET sensitivities, where for the first timewe include the intrinsic variance of the SGWB in the inference, as in this setup itpresents a dominant source of measurement noise.
Giarda et al. (Tue,) studied this question.