BriefGPT.xyz
Aug, 2018
神经形态字符串转录的模仿学习
Imitation Learning for Neural Morphological String Transduction
HTML
PDF
Peter Makarov, Simon Clematide
TL;DR
通过模仿学习，使用神经过渡型字符串转换器来进行形态学任务，如词形生成和词形归一化。采用简单的专家策略来消除对字符对齐器或热启动的需求，并在多个基准测试中实现了强大和最先进的性能。
Abstract
We employ
imitation learning
to train a
neural transition-based string transducer
for
morphological tasks
such as
→