search

floodsung / Deep-Learning-Papers-Reading-Roadmap

Deep Learning papers reading roadmap for anyone who are eager to learn this amazing tech!
38.34k stars 7.32k forks source link

Why the same model performs better in 5-way 5-shot than 5-way 1-shot ? #103

Open arcsinx11 opened 5 years ago

arcsinx11 commented 5 years ago

I really appreciate your original ideas in zhihu.When all the data can be chosen,5-shot can be better than 1-shot. 1-shot is easier than 5-shot to find the optimum weights,isn't it? So i think 1-shot can be better than 5-shot,but it not.Can you give me some advice?