一种缓解语音模型偏差的对比学习方法
A Contrastive Learning Approach to Mitigate Bias in Speech Models
AI总结：
针对语音模型在不同人群子群体中性能失衡的偏差问题，提出一种采用三级学习技术的对比学习方法，经两个口语理解数据集、两种语言验证，可改善子群体内部表征、减少偏差并提升性能。
摘要：
语音模型可能会在不同人群子群体中出现性能失衡的问题，引发了人们对这些群体间公平待遇的担忧。此前缓解不公平性的尝试要么聚焦于用户定义的子群体，可能会忽略其他受影响的子群体，要么没有在子群体层面明确改善内部表征。本文首次提出采用对比学习来缓解语音模型在表现不佳子群体中的偏差。我们采用一种三级学习技术，引导模型聚焦于对比损失的不同范围，即任务、子群体以及子群体内的误差。在两个口语理解数据集和两种语言上的实验表明，我们的方法改善了内部子群体表征，从而减少了模型偏差并提升了性能。
英文摘要：
Speech models may be affected by performance imbalance in different population subgroups, raising concerns about fair treatment across these groups. Prior attempts to mitigate unfairness either focus on user-defined subgroups, potentially overlooking other affected subgroups, or do not explicitly improve the internal representation at the subgroup level. This paper proposes the first adoption of contrastive learning to mitigate speech model bias in underperforming subgroups. We employ a three-level learning technique that guides the model in focusing on different scopes for the contrastive loss, i.e., task, subgroup, and the errors within subgroups. The experiments on two spoken language understanding datasets and two languages demonstrate that our approach improves internal subgroup representations, thus reducing model bias and enhancing performance.
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。
