Few-Shot Complex Knowledge Base Question Answering via Meta Reinforcement Learning

30141

Knowledge Triples

Source Entity	Relationship	Target Entity
Complex Question Answering Task	comparesWith	Neural Program Induction Method
Meta Reinforcement Learning Method	uses	Rewards Metric
Meta Reinforcement Learning Method	uses	Trial Trajectories OtherEntity
Neural Program Induction Method	improves	Meta Reinforcement Learning Method