Seeing Through the Shift: Causality-Inspired Robust Generalized Category Discovery
This work proposes CausalGCD, a causality-inspired framework designed to mitigate domain-shift bias in category discovery and proposes a Causal Geometric Manifold Constraint that enforces invariant manifold-level associations between known and unknown categories across domains, thereby facilitating robust discovery of novel classes.
Wei Feng, Yi-Wen Jiang, Sijin Zhou et al.
· 1 citation