Reinforced Mnemonic Reader for Machine Reading Comprehension

squad-model

Abstarct

本文介绍 the Reinforced Mnemonic Reader， 该Reader 对以前 attentive readers做了两方面的加强：
1. a reattention mechanism is proposed to refine current attentions by directly accessing to past attentions that are temporally memorized in a multi-round alignment architecture, so as to avoid the problems of attention redundancy and attention deficiency.
2. a new optimization approach, called dynamic-critical reinforcement learning, is introduced to extend the standard supervised method.

Introduction

目前的局限：
1. Tocapture complexinteractionsbetween thecontextand the question, a variety of neural attention， such as bi-attention， are proposed in a single-round alignment architecture. In order to fully compose complete information of the inputs, multiround alignment architectures that compute attentions repeatedly have been proposed。 然而， 在这些方法中，？？？
2.