TY  - RPRT
TI  - Every Model Learned by Gradient Descent Is Approximately a Kernel Machine
AU  - Pedro Domingos
PY  - 2020
UR  - https://arxiv.org/abs/2012.00152
ID  - 2012.00152
ER  -