I wasn't sure how to phrase this question, but yeah. My CNN (code here) keeps getting up to an accuracy of around 66% until it stops learning anymore. It just keeps fluctuating around the 66% range. I've been tuning hyper-parameters, but I just can't get it to work. Can anyone help?
Asked
Active
Viewed 6 times
0
-
Check the linked thread for many good hints. – Tim Aug 03 '23 at 15:48
-
i have checked that thread many-a-times. – ExplorPython Aug 03 '23 at 15:49
-
Then, if you made sure everything works fine, you either need to better tune the model, try a different model, or there is always a chance that the problem is hopeless in the sense that better performance is not possible on this data. – Tim Aug 03 '23 at 15:51
-
sounds good, i'll try it. – ExplorPython Aug 03 '23 at 15:53
-
Plus the usual suspects: improve data cleaning, preprocessing, feature engineering, get better and more data. – Tim Aug 03 '23 at 15:59