0

I have made base model for transfer learning and it is showing good accuracy, and even good confusion matrix is also showing good results

Here is accuracy and losses for base model

loss: 0.0566 - accuracy: 0.9767 - val_loss: 0.3767 - val_accuracy: 0.9613 For transfer learning model I am not able to show good accuracy but confusion matrix is excellent

Here is accuracy and loss for transfer learning

loss: 0.2698 - accuracy: 0.1914 My question is that, for transfer learning model should always less accuracy or should be more, do you think is there any problem ?

Jagdish
  • 1
  • 2
    I don't think it's possible to make general statements about transfer learning and new problems. Suppose I train an MNIST model. I don't think this model would be suitable for transfer learning if my goal were to find cancerous tumors in a sonogram. – Sycorax Nov 04 '20 at 02:33
  • 1
    The proposed duplicate is in the context of time series forecasting, but [my answer there](https://stats.stackexchange.com/a/473740/1352) makes the generally applicable point that comparing any kind of accuracy measure between problems is bound to fail, so general notions of how accuracy should behave are usually pointless. – Stephan Kolassa Nov 04 '20 at 07:23
  • Thanks I got my answer – Jagdish Nov 04 '20 at 22:21

0 Answers0