Zheng Gao

Normal Alignment is presented, a novel statistical sign-recovery approach for S1 DNNs that infers neuron signs via normal-signature alignment and delivers higher voting accuracy and pushes erroneous predictions to low-confidence ranks, which further enables a more efficient combined method, eSOE + Alignment, by combining Normal Alignment with the hard-label SOE extension.

This work introduces CoRA-NAS (COarse Ranking + Anchor-residual), a two-stage framework combining a static ranking prior with low-cost learning-curve refinement that combines cross-space ranking robustness with low-cost architecture selection.

Stochastically deflecting the reverse diffusion trajectory breaks 98–100% of generative watermarks across nine leading schemes without requiring gradient optimization, model internals, or visual quality trade-offs.

A novel algebraic approach to model extraction cuts clustering time by over 90%, enabling efficient attacks on hard-label max-pooling CNNs.