FairNVT: Improving Fairness via Noise Injection in Vision Transformers

ArXi:2604.16780v1 Announce Type: cross This paper presents FairNVT, a lightweight debiasing framework for pretrained transformer-based encoders that improves both representation and prediction level fairness while preserving task accuracy. Unlike many existing debiasing approaches that address these notions separately, we argue they are inherently connected: suppressing sensitive information at the representation level can facilitate fairer predictions.