Measure an image-classification model's performance.
Using the same images for training and evaluation would not show performance on unseen inputs.
Why should evaluation data be separated?
Data not used for training help measure how well the model generalizes to unseen inputs.
Detailed explanation
A held-out set shows whether the model works beyond the examples used to fit it.
A held-out set shows whether the model works beyond the examples used to fit it.
The split ratio depends on the project; halving storage is not the purpose.
The split ratio depends on the project; halving storage is not the purpose.
Data splitting is an evaluation design choice, not a network feature.
Data splitting is an evaluation design choice, not a network feature.
Evaluation needs reference labels; splitting does not create them.
Evaluation needs reference labels; splitting does not create them.
Try it yourself
An example you can run in a temporary verification environment.
AWS公式AIF-C01ガイドのDomain 1.3で訓練・検証・テストの役割を確認する。Expected result
訓練データの再現ではなく、未知データへの性能を測る目的を説明できる。Key points
- Generalization
- Data split
- Overfitting
Notes
- Environment: AWS公式AIF-C01試験ガイドとAWS公式ドキュメントの確認
- Command output formatting can vary slightly by distribution or tool version.
- Run the example in a temporary directory or process when possible.
Foundation review
Read the scope first
Check whether the command acts on the current shell, a new process, an existing process, or a file.
Verify the observable result
Use the supplied command and compare the output with the expected result.