Learn with Noisy Data via Unsupervised Loss Correction for Weakly Supervised Reading Comprehension

Proceedings of the 28th International Conference on Computational Linguistics, pages 2624–2634 Barcelona, Spain (Online), December 8-13, 2020 2624 Learn with Noisy Data via Unsupervised Loss Correction for Weakly Supervised Reading Comprehension Xuemiao Zhang1, Kun Zhou3, Sirui Wang4, Fuzheng Zhang4, Zhongyuan Wang4, Junfei Liu2 ∗ 1School of Software and Microelectronics, Peking University, Beijing, China 2National Engineering Research Center for Software Engineering, Peking University, Beijing, China 3Renmin University of China, Beijing, China 4Meituan-Dianping Group {zhangxuemiao,liujunfei}@pku.edu.cn,francis kun zhou@163.com {wangsirui,zhangfuzheng}@meituan.com,wzhy@outlook.com Abstract Weakly supervised machine reading comprehension (MRC) task is practical and promising for its easily available and massive training data, but inevitablely introduces noise. Existing related methods usually incorporate extra submodels to help filter noise before the noisy data is input to main models. However, these multistage methods often make training difficult, and the qualities of submodels are hard to be controlled. In this paper, we first explore and analyze the essential characteristics of noise from the perspective of loss distribution, and find that in the early stage of training, noisy samples usually lead to significantly larger loss values than clean ones. Based on the observation, we propose a hierarchical loss correction strategy to avoid fitting noise and enhance clean supervision signals, including using an unsupervisedly fitted Gaussian mixture model to calculate the weight factors for all losses to correct the loss distribution, and employ a hard bootstrapping loss to modify loss function. Experimental results on different weakly supervised MRC datasets show that the proposed methods can help improve models significantly. 1 Introduction Machine reading comprehension (MRC) (Rajpurkar et al., 2016) is a well-known NLP task, and has made significant progress in recent years (Yu et al., 2018; Devlin et al., 2019; Gong et al., 2020; Yuan et al., 2020). To learn a well-performed MRC system, large amount of human annotated data is required. However, human annotation is high-cost in real-world application, and it is hard to control the quality for some of hard instances. Recent approach (Joshi et al., 2017) utilized a distantly supervised method to collect the excerpts for answers. It greatly scales up the dataset and reduces the cost, but introduces more harmful noisy samples inevitably. There are many of approaches proposed to filter noise for question answering (QA) recently. Lin et al. (2018) and Lee et al. (2019a) adopted a paragraph selector to calculate confidences of paragraphs to help filter noisy ones before they are input into the main model. Niu et al. (2020) designed a submodel to generate labels to supervise the training of the selector. Back to MRC, Lee et al. (2019b) further proposed to generate labels for unlabeled samples, then train an extra Refinery model to refine the overall labels for multilingual MRC task with limited training data. Admittedly, these multistage methods have achieved certain improvements, but rely heavily on the se- lector, retriever or refinery. The qualities of these complementary models are hard to be controlled, and make training difficult. In fact, we can explore another novel idea that exploits the essential characteris- tics of noise …

Content truncated to 500 words