Question 24
Domain 4You are employed at a social media company and have a requirement to create a no-code image classification model for an iOS mobile application, specifically designed for identifying fashion accessories. Your labeled dataset is stored in Cloud Storage. In this context, you aim to configure a training workflow that not only minimizes cost but also provides predictions with the lowest possible latency. How should you proceed?
Correct answer: B
Explanation
Vertex AI AutoML Edge is designed for on-device deployment, which lowers inference latency and avoids repeated cloud calls. Exporting to Core ML and using the ".mlmodel" file directly fits an iOS app because Core ML runs locally on Apple devices, matching the need for the lowest possible latency and reduced serving cost.
Why each option is right or wrong
A. Train the model using Vertex AI AutoML and register the model in Vertex AI Model Registry. Configure your mobile application to send batch requests during prediction.
B. Train the model using Vertex AI AutoML Edge and export it as a Core ML model. Configure your mobile application to use the.mlmodel file directly.
Vertex AI AutoML Edge is the on-device training path intended for mobile inference, so the model can be exported for local execution rather than served from an online endpoint. For an iOS app, exporting to Core ML and embedding the .mlmodel file means predictions run on the device, which eliminates per-request online serving charges and avoids network round trips, giving the lowest latency and lowest ongoing cost for inference.
C. Train the model using Vertex AI AutoML Edge and export the model as a TFLite model. Configure your mobile application to use the.tflite file directly.
D. Train the model using Vertex AI AutoML and expose the model as a Vertex AI endpoint. Configure your mobile application to invoke the endpoint during prediction.