Kernekoncepter
Neural Architecture Search (NAS) explores flat regions in loss landscapes to optimize NN architectures for out-of-distribution (OOD) robustness.
Statistik
NAS는 NN 아키텍처를 자동으로 정의함.
FlatNAS는 OOD 강건성을 위해 NN 아키텍처를 최적화함.
SAM은 날카로운 최소화를 통해 일반화 오류를 줄임.
Citater
"FlatNAS explores flat regions in the loss landscape to optimize NN architectures for OOD robustness."
"SAM addresses overfitting by seeking flat minima through a min-max optimization problem."
"FlatNAS achieves a good trade-off between performance, OOD generalization, and the number of parameters."