TY  - RPRT
TI  - A Noise Optimum in Rehearsal-Free Continual Learning: Isolation, Mechanism, and Scope
AU  - Gunner Levi Howe
PY  - 2026
UR  - https://arxiv.org/abs/2609.20162
ID  - 2609.20162
ER  -