Feedback for 2.7: Steel plate classification

Clarity: 4/5
Quality: 5/5
Relevance: 5/5

Errors:
Suggestions:

– It’s not clear to me why the validation error in the first run was so much smaller than the training error. The fact that the test error was bad, suggests that there was indeed data leakage, no? Could that be an interesting point to mention in the presentation?

Location: AI for Materials Industry » Case 2: Defect detection on steel plates using deep learning » 2.7: Steel plate classification

Clarity: 4/5
Quality: 5/5
Relevance: 5/5

Errors:
Suggestions:

– It’s not clear to me why the validation error in the first run was so much smaller than the training error. The fact that the test error was bad, suggests that there was indeed data leakage, no? Could that be an interesting point to mention in the presentation?

  • Actions:

Leave a Reply

Your email address will not be published. Required fields are marked *