OSR: Output Space Redistribution for Adaptive Label Removal in Classification Models
A novel approach that leverages statistical redistribution in the output space to approximate the post-removal confidence vectors of a retrained model, alleviating scalability limitations and potentially mitigates privacy concerns inherent to data-dependent solutions is proposed.
Min-Yi Peng, Darian Gunamardi, Ivan Tjuawinata et al.
· 0 citations