Gradient Imitation Reinforcement Learning for Low Resource Relation Extraction
This work addresses low-resource relation extraction for natural language processing applications, presenting an incremental improvement over existing methods.
The paper tackles low-resource relation extraction by proposing a gradient imitation reinforcement learning method to address selection bias in pseudo-labeling, and introduces a framework called GradLRE that handles scenarios with or without unlabeled data. Experimental results on two public datasets show its effectiveness compared to baselines, though no concrete numbers are provided.
Low-resource Relation Extraction (LRE) aims to extract relation facts from limited labeled corpora when human annotation is scarce. Existing works either utilize self-training scheme to generate pseudo labels that will cause the gradual drift problem, or leverage meta-learning scheme which does not solicit feedback explicitly. To alleviate selection bias due to the lack of feedback loops in existing LRE learning paradigms, we developed a Gradient Imitation Reinforcement Learning method to encourage pseudo label data to imitate the gradient descent direction on labeled data and bootstrap its optimization capability through trial and error. We also propose a framework called GradLRE, which handles two major scenarios in low-resource relation extraction. Besides the scenario where unlabeled data is sufficient, GradLRE handles the situation where no unlabeled data is available, by exploiting a contextualized augmentation method to generate data. Experimental results on two public datasets demonstrate the effectiveness of GradLRE on low resource relation extraction when comparing with baselines.