Soft-Label Integration for Robust Toxicity Classification

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Cheng, Zelei, Wu, Xian, Yu, Jiahao, Han, Shuo, Cai, Xin-Qiang, Xing, Xinyu
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929583133032448
author Cheng, Zelei
Wu, Xian
Yu, Jiahao
Han, Shuo
Cai, Xin-Qiang
Xing, Xinyu
author_facet Cheng, Zelei
Wu, Xian
Yu, Jiahao
Han, Shuo
Cai, Xin-Qiang
Xing, Xinyu
contents Toxicity classification in textual content remains a significant problem. Data with labels from a single annotator fall short of capturing the diversity of human perspectives. Therefore, there is a growing need to incorporate crowdsourced annotations for training an effective toxicity classifier. Additionally, the standard approach to training a classifier using empirical risk minimization (ERM) may fail to address the potential shifts between the training set and testing set due to exploiting spurious correlations. This work introduces a novel bi-level optimization framework that integrates crowdsourced annotations with the soft-labeling technique and optimizes the soft-label weights by Group Distributionally Robust Optimization (GroupDRO) to enhance the robustness against out-of-distribution (OOD) risk. We theoretically prove the convergence of our bi-level optimization algorithm. Experimental results demonstrate that our approach outperforms existing baseline methods in terms of both average and worst-group accuracy, confirming its effectiveness in leveraging crowdsourced annotations to achieve more effective and robust toxicity classification.
format Preprint
id arxiv_https___arxiv_org_abs_2410_14894
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Soft-Label Integration for Robust Toxicity Classification
Cheng, Zelei
Wu, Xian
Yu, Jiahao
Han, Shuo
Cai, Xin-Qiang
Xing, Xinyu
Artificial Intelligence
Cryptography and Security
Machine Learning
Toxicity classification in textual content remains a significant problem. Data with labels from a single annotator fall short of capturing the diversity of human perspectives. Therefore, there is a growing need to incorporate crowdsourced annotations for training an effective toxicity classifier. Additionally, the standard approach to training a classifier using empirical risk minimization (ERM) may fail to address the potential shifts between the training set and testing set due to exploiting spurious correlations. This work introduces a novel bi-level optimization framework that integrates crowdsourced annotations with the soft-labeling technique and optimizes the soft-label weights by Group Distributionally Robust Optimization (GroupDRO) to enhance the robustness against out-of-distribution (OOD) risk. We theoretically prove the convergence of our bi-level optimization algorithm. Experimental results demonstrate that our approach outperforms existing baseline methods in terms of both average and worst-group accuracy, confirming its effectiveness in leveraging crowdsourced annotations to achieve more effective and robust toxicity classification.
title Soft-Label Integration for Robust Toxicity Classification
topic Artificial Intelligence
Cryptography and Security
Machine Learning
url https://arxiv.org/abs/2410.14894