Disentangling Causal Effects from Sets of Interventions in the Presence of Unobserved Confounders