Multi-Task Semantic Dependency Parsing with Policy Gradient for Learning Easy-First Strategies

2019/06/04 by Kurita, Shuhei, Søgaard, Anders · 1 citation
#Computation and Language (cs.CL) #FOS: Computer and information sciences

paper · doi:10.48550/arxiv.1906.01239

Abstract

In Semantic Dependency Parsing (SDP), semantic relations form directed acyclic graphs, rather than trees. We propose a new iterative predicate selection (IPS) algorithm for SDP. Our IPS algorithm combines the graph-based and transition-based parsing approaches in order to handle multiple semantic head words. We train the IPS model using a combination of multi-task learning and task-specific policy gradient training. Trained this way, IPS achieves a new state of the art on the SemEval 2015 Task 18 datasets. Furthermore, we observe that policy gradient training learns an easy-first strategy.

Cited by

• Policy Optimization Prefers The Path of Least Resistance

Related

• Deep Multitask Learning for Semantic Dependency Parsing
• Graph-Based Decoding for Task Oriented Semantic Parsing
• Transition-based Semantic Dependency Parsing with Pointer Networks
• Second-Order Semantic Dependency Parsing with End-to-End Neural Networks
• Learning compositional structures for semantic graph parsing