Validation Loss Fluctuates then Decrease alongside Validation Accuracy Increases

Question

I was working on CNN. I modified the training procedure on runtime.

As we can see from the validation loss and validation accuracy, the yellow curve does not fluctuate much. The green curve and red curve fluctuate suddenly to higher validation loss and lower validation accuracy, then goes to the lower validation loss and the higher validation accuracy, especially for the green curve.

Is it happening because of overfitting or something else?

I am asking it because, after fluctuation, the loss decreases to the lowest point, and also the accuracy increases to the highest point.

Can anyone tell me why is it happening?

Answer 1

This question has been asked a year ago when I faced this problem I searched and it hasn't any answers on it, I tried different ways and finally, data augmentation helped me. I used data augmentation and a very small learning rate. If the fluctuations are big, the batch size should be increased, and the learning rate should be decreased. After all, using more epochs helps you to have an almost smooth plot.