Metrics for evaluation of CNN

[süleyman Alemdar]

Hi everyone,

Is it possible to calculate evaluation metrics such as Precision · Recall · F1 Score for both pre-trained CNN (faultnet-unet) or user trained cnn. I want to compare the results, but I can just visually compare them. It is not enough I guess.

Thanks in advance for your answer

Sincerely

Suleyman

[Paul de Groot]

Hi Süleyman,

In our original Neural Network plugin, which is started from the Neural Networks button in the ML Control Center, we produce a statistical report for MLP classifications and you can make a confustion plot. In FaultNet and other pre-trained models we do not generate this type of output because:

1. The output of FaultNet is a Likelikood, not a classification;
2. There is no groundtruth to compare the results against.

Assuming you do have labeled data against which you want to compare output of FaultNet, or from another model, you can do so by converting the Likelihood output into a classification by using math and logic in the attribute engine to compare results.

Best regards,

Paul.

--

Paul de Groot

Special Adviser

-----------------------------------------------------------------------------------------------------------------------   dGB Earth Sciences
Phone: +31 53 4315155 E-mail: paul.d...@dgbes.com Internet: dgbes.com

----------------------------------------------------------------------------------------------------------------------- 

--
You received this message because you are subscribed to the Google Groups "OpendTect Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to users+un...@opendtect.org.
To view this discussion visit https://groups.google.com/a/opendtect.org/d/msgid/users/CADYZbsGMj2Jjx_5rwp4S%2BXMLyyr3hjgVKAxgPOOn_7DnCBRaxw%40mail.gmail.com.