Which of the following will accomplish this?
(Choose two.)
Customize the built-in image classification algorithm to use Inception and use this for model training.
Create a support case with the SageMaker team to change the default image classification algorithm to Inception.
Bundle a Docker container with TensorFlow Estimator loaded with an Inception network and use this for model training.
Use custom code in Amazon SageMaker with TensorFlow Estimator to load the model with an Inception network, and use this for model training.
Download and apt-get install the inception network code into an Amazon EC2 instance and use this instance as a Jupyter notebook in Amazon SageMaker.
Explanations:
The built-in image classification algorithm in SageMaker is not customizable to use a specific architecture like Inception directly. SageMaker only supports customizing specific hyperparameters or using a custom container.
You cannot request AWS to change the default image classification algorithm to use Inception. This option is not feasible because the SageMaker team does not offer changes to built-in algorithm architectures.
Bundling a Docker container with TensorFlow Estimator and loading the Inception network allows full flexibility in using the Inception model for image classification. This is a valid approach for training in SageMaker.
Using custom code in SageMaker with TensorFlow Estimator to load the Inception network allows for complete control over the model architecture, enabling the use of the Inception network for training.
This option requires using an EC2 instance and manually installing the necessary code, which is not an optimal or typical way to use SageMaker for training, as it doesn’t leverage SageMaker’s managed infrastructure and capabilities.