[DNN] The problem of Local Optima

yozzum·2025년 2월 2일
0

Machine Learning

목록 보기
28/30

  • (Left) In the past, people called local optima as a point where gradient is 0 when it is not global optima.
  • (Right) However, most of the time it is in fact a saddle point, not a local optima.

profile
yozzum
yozzum
이전 포스트

[DNN] Learning Rate Dacay

다음 포스트

[DNN] Hyperparametre Tuning

0개의 댓글