AI RESEARCH
Decision-Aware Quadratic ReLU Replacement for HE-Friendly Inference
arXiv CS.LG
•
ArXi:2605.22237v1 Announce Type: cross Fully homomorphic encryption (FHE) s only additions and multiplications, so FHE-only neural-network inference typically replaces ReLU with polynomials fitted over empirical activation intervals. Such interval fitting often requires higher-degree polynomials to control activation error, incurring homomorphic evaluation costs, while classification is determined by the final logit decision.