TY  - RPRT
TI  - A Non-Monotone Preconditioned Trust-Region Method for Neural Network Training
AU  - Andrea Angino
AU  - Bindi Çapriqi
AU  - Shega Likaj
AU  - Ken Trotti
AU  - Rolf Krause
PY  - 2026
UR  - https://arxiv.org/abs/2605.14860
ID  - 2605.14860
ER  -