Overfitting on the AI Learning track. Overfitting is memorizing the training rows, including their accidents. The training loss falls and the validation loss rises or stalls. The model looks brilliant in the notebook and fragile in production.
This lesson assumes you already worked through Generalization to new cases.
The idea in practice
Plot both curves. Stop when validation stops improving. Prefer a simpler model when the gap is large. Add data before you add parameters.
A concrete check
goal = {
'track': 'AI Learning',
'lesson': 'Overfitting',
}
checks = [
'input available at decision time',
'score matches the real decision',
'failure case written down',
]
print(goal['lesson'])
for item in checks:
print('-', item)
Run the sketch locally if you have Python. The printout is a reminder of the checks, not a trained model. Replace the strings with the real inputs from your own example before you treat it as a design.
What usually goes wrong
Training for more epochs because the training loss is still falling, after validation has worsened. When this happens, stop adding parameters or tools. Fix the check, the data, or the permission, then run the same example again.
What to write down
- The input you are allowed to use at decision time.
- The output and the score or pass rule.
- One failure you will test on purpose.
- What you will not claim the system can do.
Practice
Sketch two curves: healthy and overfit. Mark the epoch where you would stop.
Self-check
- Say Overfitting in one sentence that mentions an input and an output.
- Name the failure mode in this lesson and the check that would catch it.
Done when: you can explain this lesson without the page open, and you have a written failure case.