Neural network reconstruction of scalar-tensor cosmology

Neural networks have shown great promise in providing a data-first approach to exploring new physics. In this work, we use the full implementation of late time cosmological data to reconstruct a number of scalar-tensor cosmological models within the context of neural network systems. In this pipeline, we incorporate covariances in the data in the neural network training algorithm, rather than a likelihood which is the approach taken in Markov chain Monte Carlo analyses. For general subclasses of classic scalar-tensor models, we find stricter bounds on functional models which may help in the understanding of which models are observationally viable. Specifically, we show that the quintessence potential cannot deviate much from a linear behavior at the redshifts of interest, while in higher derivative theories we notice a monotonically increasing behavior for the arbitrary potentials.