i have come across few (machine learning-classification problem) journal papers mentioned evaluate accuracy top-n approach. data show top 1 accuracy = 42.5%, , top-5 accuracy = 72.5% in same training, testing condition. wonder how calculate percentage of top-1 , top-5?
can 1 show me example , steps calculate this?
thanks
top-1 accuracy conventional accuracy: model answer (the 1 highest probability) must expected answer.
top-5 accuracy means any of model 5 highest probability answers must match expected answer.
for instance, let's you're applying machine learning object recognition using neural network. picture of cat shown, , these outputs of neural network:
- tiger: 0.4
- dog: 0.3
- cat: 0.1
- lynx: 0.09
- lion: 0.08
- bird: 0.02
- bear: 0.01
using top-1 accuracy, count output wrong, because predicted tiger.
using top-5 accuracy, count output correct, because cat among top-5 guesses.
