nerdexam
Amazon

MLS-C01 · Question #284

A company is creating an application to identify, count, and classify animal images that are uploaded to the company's website. The company is using the Amazon SageMaker image classification…

The correct answer is D. Initiate transfer learning. Train the model by using the images of less common species. E. Use an augmented manifest file in JSON Lines format. To incorporate new labeled images of less common species into an existing image classification model using SageMaker Pipe mode, the ML engineer should utilize transfer learning and prepare the data as an augmented manifest file.

ML Implementation and Operations

Question

A company is creating an application to identify, count, and classify animal images that are uploaded to the company's website. The company is using the Amazon SageMaker image classification algorithm with an ImageNetV2 convolutional neural network (CNN). The solution works well for most animal images but does not recognize many animal species that are less common. The company obtains 10,000 labeled images of less common animal species and stores the images in Amazon S3. A machine learning (ML) engineer needs to incorporate the images into the model by using Pipe mode in SageMaker. Which combination of steps should the ML engineer take to train the model? (Choose two.)

Options

  • AUse a ResNet model. Initiate full training mode by initializing the network with random weights.
  • BUse an Inception model that is available with the SageMaker image classification algorithm.
  • CCreate a .lst file that contains a list of image files and corresponding class labels. Upload the .lst
  • DInitiate transfer learning. Train the model by using the images of less common species.
  • EUse an augmented manifest file in JSON Lines format.

How the community answered

(27 responses)
  • A
    7% (2)
  • B
    15% (4)
  • C
    4% (1)
  • D
    74% (20)

Why each option

To incorporate new labeled images of less common species into an existing image classification model using SageMaker Pipe mode, the ML engineer should utilize transfer learning and prepare the data as an augmented manifest file.

AUse a ResNet model. Initiate full training mode by initializing the network with random weights.

Initiating full training mode with random weights would discard the knowledge already gained by the pre-trained CNN, making the training process inefficient and requiring a much larger dataset than available for less common species.

BUse an Inception model that is available with the SageMaker image classification algorithm.

Using a different base model like Inception does not inherently address the problem of efficiently incorporating new data or leveraging existing knowledge for less common species.

CCreate a .lst file that contains a list of image files and corresponding class labels. Upload the .lst
DInitiate transfer learning. Train the model by using the images of less common species.Correct

Transfer learning is the most efficient method for adapting a pre-trained model to a new, but related, dataset, allowing the model to leverage previously learned features and quickly generalize to the less common species by fine-tuning some layers. This avoids training from scratch.

EUse an augmented manifest file in JSON Lines format.Correct

An augmented manifest file in JSON Lines format is a suitable input format for SageMaker training jobs, especially when using Pipe mode, as it allows for efficient streaming of image file paths and corresponding labels from Amazon S3 directly to the training instance for large datasets.

Concept tested: SageMaker image classification, transfer learning, Pipe mode data formats

Source: https://docs.aws.amazon.com/sagemaker/latest/dg/augmented-manifest.html

Topics

#SageMaker#Transfer Learning#Image Classification#Pipe Mode

Community Discussion

No community discussion yet for this question.

Sign up to join the discussion
Full MLS-C01 Practice