Keywords: event extraction, inverse reinforcement learning
TL;DR: We use dynamic rewards to train event extractors.
Abstract: We propose a new framework for entity and event extraction based on generative adversarial imitation learning -- an inverse reinforcement learning method using generative adversarial network (GAN). We assume that instances and labels yield to various extents of difficulty and the gains and penalties (rewards) are expected to be diverse. We utilize discriminators to estimate proper rewards according to the difference between the labels committed by ground-truth (expert) and the extractor (agent). Experiments also demonstrate that the proposed framework outperforms state-of-the-art methods.
Archival Status: Archival
Subject Areas: Natural Language Processing, Information Extraction
4 Replies
Loading