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

Media reports Hugging Face Papers Source language: Chinese

The approach aggregates within‑class partitions across diverse representation spaces to achieve cross‑view consensus, reducing reliance on spurious correlations without bias supervision and improving worst‑group accuracy.

Full text

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.

Sources

← Back to Daily AI Report