Propose an algorithm that accepts a sequence of labeled training examples and outputs a hypothesis h ∈ H that is consistent with all the training examples, if one exists. If there are no consistent hypotheses, your algorithm should indicate as such and halt