machine-learningdeep-learningoverfitting-underfitting

If the validation isnt converging, my model is overfitting right?


I have this graph, and my validation loss isn't converging. am I right that this is a case of overfitting? graph picture


Solution

  • It seems to me that this would be overfitting/high variance (big difference between loss of training set and validation set). One solution you could try would be to add some form of regularisation (L1, L2, dropout). You could also try increasing the size of your training set to get more variations in your training examples. Good luck !