Learning of networked spreading models from noisy and incomplete data