MLA-C01 Question 156
Select 4You are a Machine Learning Engineer tasked with training a deep learning model on a large dataset stored in Amazon S3. To optimize training performance, you need to ensure efficient data loading and preprocessing. Which combination of approaches should you take?
- A
Use Amazon S3 Select to filter and retrieve only the required data directly from S3.
- B
Use AWS Glue to preprocess the data and store it in Amazon DynamoDB for faster access.
- C
Use Amazon SageMaker Pipe mode to stream data directly from S3 to the training instances.
- D
Leverage Apache Spark on Amazon EMR for distributed preprocessing and store the processed data back in S3.
- E
Convert your dataset to Apache Parquet format and save it in S3 before training.
Show answer and explanation
Correct answers: A, C, D, E
Explanation
Efficient data loading and preprocessing are critical for training machine learning models on large datasets. Amazon S3 Select can optimize data retrieval by filtering data directly from S3. SageMaker Pipe mode streams data directly from S3 without requiring full dataset downloads, reducing training overhead. Apache Spark on Amazon EMR is ideal for distributed preprocessing of large datasets, and saving the data in Apache Parquet format ensures faster access and reduced storage costs. These combined approaches align with best practices for optimizing ML workflows on AWS.
- A. Correct.
Correct: Amazon S3 Select allows you to retrieve only the required data from S3, reducing the data transferred and improving efficiency.
- B. Incorrect.
Incorrect: Amazon DynamoDB is not optimized for storing or accessing large amounts of training data for machine learning models.
- C. Correct.
Correct: Amazon SageMaker Pipe mode streams data directly from S3 to the training instances, enabling efficient data loading without requiring the entire dataset to be downloaded.
- D. Correct.
Correct: Apache Spark on Amazon EMR can preprocess large datasets in a distributed manner, and storing the processed data back in S3 ensures compatibility with training processes.
- E. Correct.
Correct: Converting datasets to a columnar format like Apache Parquet improves storage efficiency and speeds up data loading during training.