Let Samples Sing in Chorus: Mitigating Spurious Correlations via Cross-view Consensus

媒体报道 Hugging Face Papers 原文语言:中文

该方法通过在多个表示空间中聚合同类划分,实现跨视图共识,从而在无需偏差监督的情况下降低对伪相关的依赖,提升最差群体的准确率。

全文

A multi-view debiasing method aggregates within-class partitions across diverse representation spaces to reduce reliance on spurious correlations and improve worst-group accuracy without bias supervision.

正文以来源语言显示。

参考来源

← 返回每日AI报告