Mohammedkaif Mohammedrafiq Kalagond

Regulating Internal Evidence Flows for Robust Learning Under Spurious Correlations

Deep models often exploit spurious correlations (e.g., backgrounds or dataset artifacts), hurting worst-group performance. We propose \textbf{Evidence-Gated Suppression (EGS)}, a lightweight, plug-in regularizer that intervenes inside the network during training. EGS tracks a class-conditional, conf…