arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2205.01588cs.CLcs.AI
收藏

SparCAssist:一个基于稀疏反事实生成的风险评估模型助手

SparCAssist: A Model Risk Assessment Assistant Based on Sparse Generated Counterfactuals

L3S Research Center(L3S研究中心);University of Stavanger(斯塔万格大学);Delft University of Technology(代尔夫特理工大学)

Zijian Zhang, Vinay Setty, Avishek Anand

摘要:

我们介绍SparcAssist,一个通用的风险评估工具,用于为语言任务训练的机器学习模型。它通过检查模型在反事实上的行为来评估模型的风险,即基于给定数据实例生成的分布外实例。反事实是通过替换由ExPred识别的理性一致性中的标记来生成的,而替换是使用HotFlip或基于Masked-Language-Model的算法来检索的。我们的系统的主要目的是帮助人类注释者评估模型部署的风险。在评估过程中生成的反事实实例是副产品,将来可以用来训练更健壮的NLP模型。

英文摘要:

We introduce SparcAssist, a general-purpose risk assessment tool for the machine learning models trained for language tasks. It evaluates models' risk by inspecting their behavior on counterfactuals, namely out-of-distribution instances generated based on the given data instance. The counterfactuals are generated by replacing tokens in rational subsequences identified by ExPred, while the replacements are retrieved using HotFlip or Masked-Language-Model-based algorithms. The main purpose of our system is to help the human annotators to assess the model's risk on deployment. The counterfactual instances generated during the assessment are the by-product and can be used to train more robust NLP models in the future.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑