0

I have a supervised images classification problem, I am using Convolutional neural network model to solve it.

there is 8 classes:

what can be result in good accuracy to train the model on all the 8 classes, or to divide them into 2 models with 4 classes each;

8 classes Vs 4 classes in supervised classification

to summarise: which is better to train a model on lot of classes or few classes ?

Does a lot of classes returns a better accuracy ?

Guissous Allaeddine
  • 101
  • 3
  • 2
    You can't really compare the accuracy for the four class problem to the accuracy for the eight class problem, they are simply different problem contexts. Whether it is a good idea to group your eight classes into four depends on what problem you are trying to solve. It is generally easier to seperate four classes than eight, all else being equal, but is distinguishing the four classes sufficient to solve your problem? – Matthew Drury Aug 08 '19 at 16:53
  • Recommended reading: [Why is accuracy not the best measure for assessing classification models?](https://stats.stackexchange.com/q/312780/1352) [Is accuracy an improper scoring rule in a binary classification setting?](https://stats.stackexchange.com/q/359909/1352) [Classification probability threshold](https://stats.stackexchange.com/q/312119/1352) – Stephan Kolassa Aug 09 '19 at 17:10

0 Answers0