But there have not been rigorous evaluations showing for exactly which tasks this syntax-based method is appropriate. In this paper we benchmark {\bf recursive} neural models against sequential {\bf recurrent} neural models (simple recurrent and LSTM models). We investigate 4 tasks: (1) sentiment classification at the sentence level and phrase level; (2) matching questions to answer-phrases; (3) discourse parsing; (4) semantic relation extraction (e.g., {\em component-whole} between nouns). We implement carefully matched versions of recursive and recurrent models and apply them to each task.
Our analysis suggests that sequence-based recurrent models achieve equal or better performance in most tasks with the exception that syntactic tree-based recursive models are particularly helpful for tasks that require representing long-distance relations between words (e.g., semantic relations between nominals). Our results help understand the limitations of both classes of models, and suggest directions for improving recurrent models \footnote{Code for models described in Section 3.1 on Stanford Treebank dataset are released at\url{this http URL}.}.
from cs.AI updates on arXiv.org http://ift.tt/1B4ICML
via IFTTT
No comments:
Post a Comment