DCN+: Mixed objective and deep residual coattention for question answering
Caiming Xiong, Victor Zhong, Richard Socher Salesforce Research Palo Alto, CA 94301, USA {cxiong, vzhong, Equal contribution
Abstract
Traditional models for question answering optimize using cross entropy loss, which encourages exact answers at the cost of penalizing nearby or overlapping answers that are sometimes equally accurate. We propose a mixed objective that combines cross entropy loss with self-critical policy learning. The objective uses rewards derived from word overlap to solve the misalignment between evaluation metric and optimization objective. In addition to the mixed objective, we improve dynamic coattention networks (DCN) with a deep residual coattention encoder that is inspired by recent work in deep self-attention and residual networks. Our proposals improve model performance across question types and input lengths, especially for long questions that requires the ability to capture long-term dependencies. On the Stanford Question Answering Dataset, our model achieves state-of-the-art results with 75.1% exact match accuracy and 83.1% F1, while the ensemble obtains 78.9% exact match accuracy and 86.0% F1.
中文速览
机器阅读理解模型在训练时通常用交叉熵损失来定位答案的起止位置，但这会导致一个问题：与正确答案文字高度重叠却位置略有偏差的预测，会被惩罚得和完全错误的预测一样重，训练目标与F1评估指标严重脱节。为此，作者提出将交叉熵损失与基于词语重叠奖励的自我批评强化学习（self-critical policy learning）相结合的混合目标函数，同时对动态协同注意力网络（Dynamic Coattention Network, DCN）的编码器进行改造，引入多层残差协同注意力结构，使模型能更好地捕捉长距离依赖关系。在斯坦福问答数据集（SQuAD）上，改进后的模型DCN+单模型达到75.1%精确匹配和83.1% F1，集成后进一步提升至78.9%精确匹配和86.0% F1，刷新了当时的最优成绩。这项工作的价值在于同时从训练目标对齐和网络结构深化两个维度解决了问题，为后续问答系统的优化提供了可借鉴的思路。
原文 arXiv:1711.00106;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1711.00106v2