深度学习调参建议

1. Grid Search v.s. Random Search

Assumption: top K results are good enough.

if there are N points, probability K/N that your sample is in top K Sample x times: [
1 - (1 - K /N )^x > 90 %
] If N = 1000, K = 10 ———–-> x = 230

​ K = 100 ————> x = 22

2. Model-based Hyperparameter

3. learn to learn