Entering edit mode
3.9 years ago
shome
▴
10
I am currently trying to build a classifier using 1-D CNN using data that I obtained from protein structures. I am facing the accuracy of the model is around 38 percent, and I have experimented with different combinations of hyperparameters. It hasn't improved a lot(the best accuracy is still at 38 percent). I included more data, but still not much improvement.How to approach this issue ?
What is the number of datasets and classes? please mention train, test, and validation percentage?