Regularization on the AI Learning track. Regularization punishes complexity so the model does not fit noise. Weight decay, dropout, and limits on tree depth are common. It is a bet that smoother functions will generalize.
This lesson assumes you already worked through Batches.
The idea in practice
Add regularization when the train-validation gap is large. Do not add it so strongly that both scores collapse.
A concrete check
goal = {
'track': 'AI Learning',
'lesson': 'Regularization',
}
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
Dropout on a tiny dataset with a tiny model can be the reason nothing learns. 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
Name the symptom that would make you add weight decay.
Self-check
- Say Regularization 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.