MLA-C01 Question Bank · Latest 50 Questions Free Preview (Page 3/5)
Questions 21–30 of the latest 50 MLA-C01 questions (5 pages total), with answers and explanations, no login required. Practice all 233 questions online.
Question 21 · No. 204 · Single choice
An ML engineer needs to develop an AI assistant by using Retrieval Augmented Generation (RAG) in Amazon Bedrock. The company stores a collection of PDF text files in an Amazon S3 bucket. The ML engineer must create an Amazon Bedrock knowledge base to process the PDF files and to store the processed files in a vector store. Which solution will meet these requirements?
- A. Use the latest version of Amazon Titan Text Premier to perform document embedding.
- B. Use the latest version of Mistral 7B Instruct to perform instruction tuning.
- C. Use the latest version of Anthropic Claude Sonnet to perform instruction tuning.
- D. Use Cohere Embed Multilingual to perform document embedding.
Answer: D
Explanation: For creating an Amazon Bedrock knowledge base with RAG, an embedding model is required to convert PDF text files into vector representations that can be stored in a vector store. Option D (Cohere Embed Multilingual) is specifically designed for document embedding, which is the correct use case here. Options A (Amazon Titan Text Premier), B (Mistral 7B Instruct), and C (Anthropic Claude Sonnet) are all text generation/conversational models, not embedding models, so they cannot be used to create vector embeddings for the knowledge base.
Question 22 · No. 205 · Single choice
An ML engineer is tuning an image classification model that shows poor performance on one of two available classes during prediction. Analysis reveals that the images whose class the model performed poorly on represent an extremely small fraction of the whole training dataset. The ML engineer must improve the model’s performance. Which solution will meet this requirement?
- A. Optimize for accuracy. Use image augmentation on the less common images to generate new samples.
- B. Optimize for F1 score. Use image augmentation on the less common images to generate new samples.
- C. Optimize for accuracy. Use Synthetic Minority Oversampling Technique (SMOTE) on the less common images to generate new samples.
- D. Optimize for F1 score. Use Synthetic Minority Oversampling Technique (SMOTE) on the less common images to generate new samples.
Answer: B
Explanation: When dealing with imbalanced image classification, two key considerations apply: (1) SMOTE is designed for tabular/numerical features and works by interpolating between samples in feature space, which would destroy the spatial structure of images and produce nonsensical results - image augmentation (rotation, flipping, cropping, etc.) is the correct approach for image data. (2) Accuracy is a poor metric for imbalanced datasets because a model can achieve high accuracy by simply predicting the majority class. The F1 score is more appropriate as it balances precision and recall, giving a better picture of performance on the minority class. Therefore, optimizing for F1 score combined with image augmentation on the minority class is the correct solution.
Question 23 · No. 206 · Single choice
An ML engineer at an entertainment company is refining an ML model to predict audience preferences. As the model learns from data, the ML engineer notices that the model’s performance peaks early and then begins to gradually decline. The ML engineer must prevent the performance degradation after initial success. Which solution will meet this requirement?
- A. Increase the number of layers.
- B. Implement early stopping.
- C. Add more neurons to each layer to capture complex patterns.
- D. Examine model bias and variance to understand performance issues.
Answer: B
Explanation: The scenario describes a model whose performance peaks early and then gradually declines, which is a classic symptom of overfitting. The model initially learns the underlying patterns but then begins to memorize noise in the training data, causing generalization performance to degrade. Early stopping is a regularization technique specifically designed to address this issue—it monitors the model's performance on a validation set during training and halts the training process when performance begins to decline, thereby preserving the best-performing model state. Option A (adding layers) and Option C (adding neurons) would increase model complexity and likely worsen overfitting. Option D (examining bias and variance) is a diagnostic activity rather than a concrete solution to prevent the degradation.
Question 24 · No. 207 · Single choice
An ML engineer decides to use Amazon SageMaker AI automated model tuning (AMT) for hyperparameter optimization (HPO). The ML engineer requires a tuning strategy that uses regression to slowly and sequentially select the next set of hyperparameters based on previous runs. The strategy must work across small hyperparameter ranges. Which solution will meet these requirements?
- A. Grid search
- B. Random search
- C. Bayesian optimization
- D. Hyperband
Answer: C
Explanation: Bayesian optimization uses regression models to evaluate previous tuning runs and sequentially select the next hyperparameter combinations. It is well suited to optimization within small hyperparameter ranges.
Question 25 · No. 208 · Single choice
An ML engineer is developing a linear regression ML model. The model shows high accuracy on the training dataset but performs poorly on unseen new data. Which action should the ML engineer take to address this issue?
- A. Increase the complexity of the model to capture more patterns in the training data. Use Amazon SageMaker Debugger to monitor for convergence issues.
- B. Apply ML techniques such as cross-validation and regularization. Use Amazon SageMaker Experiments to track and compare different model versions and their performance metrics.
- C. Directly deploy the model into production. Use Amazon SageMaker Clarify to interpret model outputs on new data. Adjust the model based on these insights.
- D. Increase the size of the training dataset without adjusting the size of the model. Retrain the model on the new data. Generate a confusion matrix to analyze the results.
Answer: B
Explanation: The model is overfitting: high accuracy on training data but poor performance on unseen data. Option B correctly addresses this by recommending cross-validation (to detect overfitting) and regularization (to prevent overfitting by penalizing complex models). Amazon SageMaker Experiments is also appropriate for tracking and comparing different model versions. Option A would worsen overfitting by increasing complexity. Option C ignores the overfitting problem entirely. Option D only partially helps and includes a confusion matrix, which is irrelevant for regression models.
Question 26 · No. 209 · Single choice
A company is using an ML model to classify motion in videos. The data is stored in MP4 format in Amazon S3. When the company created the model, the company needed 4 months to label all the video frames. The company needs to retrain the model with an existing training workflow in Amazon SageMaker AI. An ML engineer must implement a solution that decreases the labeling time. Which solution will meet these requirements?
- A. Use SageMaker Ground Truth to annotate the video frames.
- B. Use SageMaker JumpStart to use pre-trained computer vision models to develop a labeling model.
- C. Use SageMaker Data Wrangler to create a data workflow. Use the workflow to optimize the labeling process.
- D. Use the labeling interface of Amazon Augmented AI (Amazon A2I) with Amazon Rekognition to label the video frames.
Answer: A
Explanation: SageMaker Ground Truth is the appropriate service for annotating video frames. It supports video frame labeling and offers automated labeling capabilities, where ML models pre-label data and humans verify the labels. This significantly reduces labeling time compared to fully manual labeling. Option B (JumpStart) provides pre-trained models for model development, not labeling. Option C (Data Wrangler) is for data preparation, not labeling. Option D (A2I with Rekognition) is for human review of ML predictions, not for initial dataset labeling.
Question 27 · No. 210 · Single choice
A company uses an ML model to recommend videos to users. The model is deployed on Amazon SageMaker AI. The model performed well initially after deployment, but the model’s performance has degraded over time. Which solution can the company use to identify model drift in the future?
- A. Create a monitoring job in SageMaker Model Monitor. Then create a baseline from the training dataset.
- B. Create a baseline from the training dataset. Then create a monitoring job in SageMaker Model Monitor.
- C. Create a baseline by using a built-in rule in SageMaker Clarify. Monitor the drift in Amazon CloudWatch.
- D. Retrain the model on new data. Compare the retrained model's performance to the original model’s performance.
Answer: B
Explanation: The correct workflow for using SageMaker Model Monitor to detect model drift is: first create a baseline from the training dataset (which serves as the reference for comparison), and then create a monitoring job that compares live inference data against this baseline to detect any drift over time. Option A has the order reversed. Option C incorrectly uses SageMaker Clarify, which is designed for bias detection, not model drift monitoring. Option D describes a retraining approach, not a monitoring solution for detecting drift.
Question 28 · No. 211 · Single choice
An ML engineer used Amazon SageMaker Studio to train a neural network. The neural network logs its information into TensorBoard and uses stochastic gradient descent (SGD) as the optimizer. The ML engineer reviewed training graphs and discovered that the accuracy was not increasing and the loss was decreasing very slowly. The ML engineer needs to improve the model’s performance without increasing the total training time. Which solution will meet these requirements?
- A. Increase the initial learning rate.
- B. Decrease the initial learning rate.
- C. Increase the total number of epochs.
- D. Decrease the total number of epochs.
Answer: A
Explanation: When using SGD optimizer, if the accuracy is not increasing and the loss is decreasing very slowly, this typically indicates that the learning rate is too low. The model is making very small updates to the weights, resulting in slow convergence. Increasing the initial learning rate will allow the model to make larger weight updates, helping it converge faster and improve accuracy without requiring additional training time. Option B would worsen the problem, while options C and D would either increase training time or potentially harm model performance.
Question 29 · No. 212 · Multiple choice
An ML engineer is importing a custom model from the Hugging Face Hub into Amazon Bedrock. The ML engineer wants to use the model with the Amazon Bedrock API in an agentic AI application. Which combination of steps will meet these requirements? (Choose two.)
- A. Import the model for deployment by using Amazon Bedrock On-Demand mode.
- B. Convert the model to a quantized format. Import the quantized model after conversion.
- C. Import the model directly by using the Amazon SageMaker AI SDK in the agentic AI application.
- D. During the model import process, configure an auto-scaling policy before using the model in the agentic AI application.
- E. In the import job, provide the complete Amazon S3 URI that points to the Hugging Face model files. Ensure that the S3 URI is in the same AWS account as the Amazon Bedrock import job.
Answer: A, E
Question 30 · No. 213 · Single choice
A digital media entertainment company needs real-time video content moderation to ensure compliance during live streaming events. Which solution will meet these requirements with the LEAST operational overhead?
- A. Use Amazon Rekognition and AWS Lambda to extract and analyze the metadata from the videos’ image frames.
- B. Use Amazon Rekognition and a large language model (LLM) hosted on Amazon Bedrock to extract and analyze the metadata from the videos’ image frames.
- C. Use Amazon SageMaker AI to extract and analyze the metadata from the videos’ image frames.
- D. Use Amazon Transcribe and Amazon Comprehend to extract and analyze the metadata from the videos’ image frames.
Answer: A
Explanation: Amazon Rekognition has a built-in content moderation feature specifically designed to detect inappropriate visual content in images and videos, and it can process video streams in real-time. Combined with AWS Lambda (a serverless compute service), this solution requires no infrastructure management, no model training, and minimal operational overhead. Option B adds unnecessary complexity by using an LLM on Bedrock when Rekognition already handles image analysis natively. Option C (SageMaker AI) requires significant operational overhead for building, training, and maintaining custom ML models. Option D (Transcribe and Comprehend) is designed for audio and text analysis, not video image content moderation.