1
$\begingroup$

I've been statistically validating the performance of different Deep Learning models in classifying parasitized and normal cells. In the process, I could find that there is no statistically significant difference in performance between the pre-trained CNN models. In the absence of such significance, which model am I supposed to choose?

$\endgroup$
0
$\begingroup$

This is a very general question, since you are not providing any statistics, that one can comment on. But in general: yes, this can be the case. However, there is an interesting paper on comparing classification models with statistical tests written by Demšar (2006). He shows a way to vizualize the results of different models that do not significantly differ in quality.

$\endgroup$
  • $\begingroup$ This method operates on average ranks, which requires several datasets, but it seems like OP only has one dataset to evaluate on. $\endgroup$ – timleathart Jan 4 '18 at 22:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.