TY  - RPRT
TI  - Physical Neural Networks Need Nonlinearity, Amplification, and Suppression for Learning
AU  - Nex Chiaki Xijana Stuhlmüller
AU  - Marjolein Dijkstra
PY  - 2026
UR  - https://arxiv.org/abs/2606.26989
ID  - 2606.26989
ER  -