A model overfits. Which change helps?
Why A is the answer
Add regularization or more training data
Regularization and more data reduce overfitting.
In detail
- Regularization or more data reduces overfitting.
- Either change can help on its own.
Use case
An illustrative scenario, not a real incident.
A model memorizes its small dataset. Regularization and more data both push it toward general patterns.
See it in different ways
flowchart TD
A{"Results"} -->|"Poor on training"| B["Underfitting: more features or a bigger model"]
A -->|"Good on training, poor on validation"| C["Overfitting: regularize or add data"]stateDiagram-v2 [*] --> Train Train --> Validate: tune Validate --> Train: adjust Validate --> Test: choose the final model Test --> [*]
mindmap
root(("Overfitting fixes"))
n1["Model"]
n2["Regularization"]
n3["Data"]
n4["More training data"]Key point: Regularization and more data reduce overfitting.
Why the other options are wrong
- B. Remove the validation set from the process
- Without validation, you cannot see it.
- C. Add more layers and train for longer
- More layers can make it worse.
- D. Stop evaluating on data it has not seen
- Unseen data is how you detect it.