Google Professional Machine Learning Engineer

• #141

  You aim to redesign your ML pipeline for structured data on Google Cloud. Currently, you employ PySpark for large-scale data transformations, but your pipelines take over 12 hours to run. To accelerate development and pipeline execution, you intend to leverage a serverless tool and SQL syntax. With your raw data already migrated to Cloud Storage, how should you construct the pipeline on Google Cloud to meet the speed and processing requirements?

  • AA. Use Data Fusion's GUI to construct the transformation pipelines, and then write the data into BigQuery.
  • BB. Convert your PySpark code into SparkSQL queries for data transformation, and subsequently, execute your pipeline on Dataproc to write the data into BigQuery.
  • CC. Ingest your data into Cloud SQL, convert your PySpark commands into SQL queries to transform the data, and then employ federated queries from BigQuery for machine learning.
  • DD. Ingest your data into BigQuery using BigQuery Load, convert your PySpark commands into BigQuery SQL queries for data transformation, and then save the transformations to a new table.
  Show answer
• #142

  You are developing an ML pipeline using Vertex AI Pipelines. You want your pipeline to upload a new version of the XGBoost model to Vertex AI Model Registry and deploy it to Vertex AI Endpoints for online inference. You want to use the simplest approach. What should you do?

  • AA. Chain the Vertex AI ModelUploadOp and ModelDeployOp components together.
  • BB. Use the Vertex AI ModelEvaluationOp component to evaluate the model.
  • CC. Use the Vertex AI SDK for Python within a custom component based on a python:3.10 image.
  • DD. Use the Vertex AI REST API within a custom component based on a vertex-ai/prediction/xgboost-cpu image.
  Show answer
• #143

  You are employed at a retail company and have developed a Vertex AI forecast model that produces monthly item sales predictions. Now, you aim to swiftly generate a report that explains how the model calculates these predictions. You possess one month of recent actual sales data that was not part of the training dataset. What steps should you take to generate data for your report?

  • AA. Create a batch prediction job using the actual sales data and compare the predictions to the actuals in the report.
  • BB. Create a batch prediction job using the actual sales data and configure the job settings to generate feature attributions. Then, compare the results in the report.
  • CC. Generate counterfactual examples using the actual sales data. Subsequently, create a batch prediction job by using both the actual sales data and the counterfactual examples, and compare the results in the report.
  • DD. Train another model using the same training dataset as the original, excluding some columns. Then, utilize the actual sales data to create one batch prediction job with the new model and another with the original model. Finally, compare the two sets of predictions in the report.
  Show answer
• #144

  You are creating a model training pipeline to predict sentiment scores from text-based product reviews. You want to have control over how the model parameters are tuned, and you will deploy the model to an endpoint after it has been trained. You will use Vertex AI Pipelines to run the pipeline. You need to decide which Google Cloud pipeline components to use. What components should you choose?

  • AA. TabularDatasetCreateOp, CustomTrainingJobOp, and EndpointCreateOp
  • BB. TextDatasetCreateOp, AutoMLTextTrainingOp, and EndpointCreateOp
  • CC. TabularDatasetCreateOp, AutoMLTextTrainingOp, and ModelDeployOp
  • DD. TextDatasetCreateOp, CustomTrainingJobOp, and ModelDeployOp
  Show answer
• #145

  You are an ML engineer at a global car manufacturer. Your task is to develop an ML model for predicting car sales in various cities worldwide. Which features or feature combinations should you use to capture city-specific relationships between car types and the number of sales?

  • AA. Three individual features: binned latitude, binned longitude, and one-hot encoded car type.
  • BB. One feature obtained as an element-wise product between latitude, longitude, and car type.
  • CC. One feature obtained as an element-wise product between binned latitude, binned longitude, and one-hot encoded car type.
  • DD. Two feature crosses as an element-wise product: the first between binned latitude and one-hot encoded car type, and the second between binned longitude and one-hot encoded car type.
  Show answer
• #146

  You have recently developed the initial version of an image segmentation model for a self-driving car. Upon deploying the model, you notice a decline in the area under the curve (AUC) metric. Additionally, upon reviewing video recordings, you find that the model performs poorly in densely congested traffic scenarios but functions correctly in lower-traffic situations. What is the most probable explanation for this outcome?

  • AA. The model is overfitting in areas with less traffic and underfitting in areas with more traffic.
  • BB. AUC is not the correct metric to evaluate this classification model.
  • CC. Too much data representing congested areas was used for model training.
  • DD. Gradients become small and vanish while backpropagating from the output to input nodes.
  Show answer
• #147

  You're working on an ML model within a Vertex AI Workbench notebook and aim to track artifacts and compare models during experimentation while efficiently transitioning successful experiments to production as you iterate on your model implementation. What should you do?

  • AA. 1. Initialize the Vertex SDK with the experiment name. Log parameters and metrics for each experiment, and attach dataset and model artifacts as inputs and outputs to each execution. 2. After a successful experiment, create a Vertex AI pipeline.
  • BB. 1. Initialize the Vertex SDK with the experiment name. Log parameters and metrics for each experiment, save your dataset to a Cloud Storage bucket, and upload the models to Vertex AI Model Registry. 2. After a successful experiment, create a Vertex AI pipeline.
  • CC. 1. Create a Vertex AI pipeline with parameters tracked as arguments to your PipelineJob. Utilize Metrics, Model, and Dataset artifact types from the Kubeflow Pipelines DSL as inputs and outputs of the pipeline components. 2. Associate the pipeline with your experiment upon job submission.
  • DD. 1. Create a Vertex AI pipeline. Use Dataset and Model artifact types from the Kubeflow Pipelines DSL as inputs and outputs of the pipeline components. 2. Within your training component, employ the Vertex AI SDK to create an experiment run. Configure the log_params and log_metrics functions to track parameters and metrics of your experiment.
  Show answer
• #148

  You have effectively deployed a substantial and intricate TensorFlow model that was trained on tabular data. Your objective is to predict the lifetime value (LTV) field for each subscription, which is stored in the BigQuery table named "subscription.subscriptionPurchase" within the "my-fortune500-company-project" project. To ensure that prediction drift is prevented, which refers to significant changes in feature data distribution in production over time, what steps should you take?

  • AA. Implement continuous retraining of the model daily using Vertex AI Pipelines.
  • BB. Add a model monitoring job where 10% of incoming predictions are sampled 24 hours.
  • CC. Add a model monitoring job where 90% of incoming predictions are sampled 24 hours.
  • DD. Add a model monitoring job where 10% of incoming predictions are sampled every hour.
  Show answer
• #149

  You work for an advertising company and aim to evaluate the effectiveness of your latest advertising campaign. You've streamed 500 MB of campaign data into BigQuery and want to query the table, followed by manipulating the query results using a pandas dataframe in an AI Platform notebook. What's the recommended approach?

  • AA. Use AI Platform Notebooks' BigQuery cell magic to query the data, and ingest the results as a pandas dataframe.
  • BB. Export your table as a CSV file from BigQuery to Google Drive, and use the Google Drive API to ingest the file into your notebook instance.
  • CC. Download your table from BigQuery as a local CSV file, and upload it to your AI Platform notebook instance. Use pandas.read_csv to ingest the file as a pandas dataframe.
  • DD. From a bash cell in your AI Platform notebook, use the bq extract command to export the table as a CSV file to Cloud Storage, and then use gsutil cp to copy the data into the notebook. Use pandas.read_csv to ingest the file as a pandas dataframe.
  Show answer
• #150

  You are tasked with creating a custom deep neural network in Keras to forecast customer purchases based on their purchase history. To assess the performance across various model architectures, while storing training data and comparing evaluation metrics on a unified dashboard, what approach should you adopt?

  • AA. Create multiple models using Vertex AI AutoML Tables.
  • BB. Automate multiple training runs using Cloud Composer.
  • CC. Run multiple training jobs on AI Platform with similar job names.
  • DD. Create an experiment in Kubeflow Pipelines to organize multiple runs.
  Show answer
• #151

  You are part of a food product company, and your historical sales data is stored in BigQuery. Your task is to utilize Vertex AI's custom training service to train multiple TensorFlow models, leveraging the data from BigQuery to predict future sales. In preparation for model experimentation, you plan to implement a data preprocessing algorithm that involves min-max scaling and bucketing for a significant number of features. Your aim is to keep preprocessing time, costs, and development efforts to a minimum. How should you configure this workflow?

  • AA. Write the transformations into Spark that uses the spark-bigquery-connector, and use Dataproc to preprocess the data.
  • BB. Write SQL queries to transform the data in-place in BigQuery.
  • CC. Add the transformations as a preprocessing layer in the TensorFlow models.
  • DD. Create a Dataflow pipeline that uses the BigQuery lO connector to ingest the data, process it, and write it back to BigQuery.
  Show answer
• #152

  You recently developed a custom ML model that was trained in Vertex AI on a post-processed training dataset stored in BigQuery. You used a Cloud Run container to deploy the prediction service. The service performs feature lookup and pre-processing and sends a prediction request to a model endpoint in Vertex AI. You want to configure a comprehensive monitoring solution for training-serving skew that requires minimal maintenance. What should you do?

  • AA. Create a Model Monitoring job for the Vertex AI endpoint that uses the training data in BigQuery to perform training-serving skew detection and uses email to send alerts. When an alert is received, use the console to diagnose the issue.
  • BB. Update the model hosted in Vertex AI to enable request-response logging. Create a Data Studio dashboard that compares training data and logged data for potential training-serving skew and uses email to send a daily scheduled report.
  • CC. Create a Model Monitoring job for the Vertex AI endpoint that uses the training data in BigQuery to perform training-serving skew detection and uses Cloud Logging to send alerts. Set up a Cloud Function to initiate model retraining that is triggered when an alert is logged.
  • DD. Update the model hosted in Vertex AI to enable request-response logging. Schedule a daily DataFlow Flex job that uses Tensorflow Data Validation to detect training-serving skew and uses Cloud Logging to send alerts. Set up a Cloud Function to initiate model retraining that is triggered when an alert is logged.
  Show answer
• #153

  You've utilized Vertex AI Workbench notebooks to construct a TensorFlow model, and the notebook follows these steps: 1. Fetching data from Cloud Storage, 2. Employing TensorFlow Transform for data preprocessing, 3. Utilizing native TensorFlow operators to define a sequential Keras model, 4. Conducting model training and evaluation using model.fit() within the notebook instance, and 5. Storing the trained model in Cloud Storage for serving. Your objective is to orchestrate a weekly model retraining pipeline with minimal cost, refactoring, and monitoring efforts. How should you proceed to achieve this?

  • AA. Add relevant parameters to the notebook cells and set a recurring run in Vertex AI Workbench.
  • BB. Use TensorFlow Extended (TFX) with Google Cloud executors to define your pipeline, and automate the pipeline to run on Cloud Composer.
  • CC. Use Kubeflow Pipelines SDK with Google Cloud executors to define your pipeline, and use Cloud Scheduler to automate the pipeline to run on Vertex AI Pipelines.
  • DD. Use TensorFlow Extended (TFX) with Google Cloud executors to define your pipeline, and use Cloud Scheduler to automate the pipeline to run on Vertex AI Pipelines.
  Show answer
• #154

  You are in the process of deploying a new version of a model to a production Vertex AI endpoint that is actively serving user traffic. Your goal is to direct all user traffic to the new model while minimizing any disruption to your application. How should you proceed to achieve this objective?

  • AA. Create a new endpoint Create a new model. Set it as the default version. Upload the model to Vertex AI Model Registry Deploy the new model to the new endpoint Update Cloud DNS to point to the new endpoint
  • BB. Create a new endpoint Create a new model. Set the parentModel parameter to the model ID of the currently deployed model and set it as the default version. Upload the model to Vertex AI Model Registry Deploy the new model to the new endpoint, and set the new model to 100% of the traffic.
  • CC. Create a new model. Set the parentModel parameter to the model ID of the currently deployed model. Upload the model to Vertex AI Model Registry. Deploy the new model to the existing endpoint, and set the new model to 100% of the traffic
  • DD. Create a new model. Set it as the default version. Upload the model to Vertex AI Model Registry Deploy the new model to the existing endpoint
  Show answer
• #155

  You're creating a Kubeflow pipeline on Google Kubernetes Engine, where the initial step involves querying BigQuery. The query results will serve as input for the subsequent step in your pipeline. What is the simplest method to accomplish this?

  • AA. Use the BigQuery console to execute your query, and then save the query results into a new BigQuery table.
  • BB. Write a Python script that uses the BigQuery API to execute queries against BigQuery. Execute this script as the first step in your Kubeflow pipeline.
  • CC. Use the Kubeflow Pipelines domain-specific language to create a custom component that uses the Python BigQuery client library to execute queries.
  • DD. Locate the Kubeflow Pipelines repository on GitHub. Find the BigQuery Query Component, copy that component's URL, and use it to load the component into your pipeline. Use the component to execute queries against BigQuery.
  Show answer
• #156

  Your team is currently engaged in an NLP research project aimed at predicting the political affiliations of authors based on the articles they have authored. The training dataset for this project is extensive and structured as follows: AuthorA:Political Party A TextA1: [SentenceA11, SentenceA12, SentenceA13, ...] TextA2: [SentenceA21, SentenceA22, SentenceA23, ...] … AuthorB:Political Party B TextB1: [SentenceB11, SentenceB12, SentenceB13, ...] TextB2: [SentenceB21, SentenceB22, SentenceB23, ...] … AuthorC:Political Party B TextC1: [SentenceC11, SentenceC12, SentenceC13, ...] TextC2: [SentenceC21, SentenceC22, SentenceC23, ...] … AuthorD:Political Party A TextD1: [SentenceD11, SentenceD12, SentenceD13, ...] TextD2: [SentenceD21, SentenceD22, SentenceD23, ...] … … To maintain the standard 80%-10%-10% data distribution across the training, testing, and evaluation subsets, you should distribute the training examples as follows:

  • AA. Distribute texts randomly across the train-test-eval subsets: Train set: [TextA1, TextB2, ...] Test set: [TextA2, TextC1, TextD2, ...] Eval set: [TextB1, TextC2, TextD1, ...]
  • BB. Distribute authors randomly across the train-test-eval subsets: Train set: [TextA1, TextA2, TextD1, TextD2, ...] Test set: [TextB1, TextB2, ...] Eval set: [TextC1, TextC2, ...]
  • CC. Distribute sentences randomly across the train-test-eval subsets: Train set: [SentenceA11, SentenceA21, SentenceB11, SentenceB21, SentenceC11, SentenceD21, ...] Test set: [SentenceA12, SentenceA22, SentenceB12, SentenceC22, SentenceC12, SentenceD22, ...] Eval set: [SentenceA13, SentenceA23, SentenceB13, SentenceC23, SentenceC13, SentenceD31, ...]
  • DD. Distribute paragraphs of texts (i.e., chunks of consecutive sentences) across the train-test-eval subsets: Train set: [SentenceA11, SentenceA12, SentenceD11, SentenceD12, ...] Test set: [SentenceA13, SentenceB13, SentenceB21, SentenceD23, SentenceC12, SentenceD13, ...] Eval set: [SentenceA11, SentenceA22, SentenceB13, SentenceD22, SentenceC23, SentenceD11, ...]
  Show answer
• #157

  Your data science team needs to rapidly experiment with various features, model architectures, and hyperparameters. They need to track the accuracy metrics for various experiments and use an API to query the metrics over time. What should they use to track and report their experiments while minimizing manual effort?

  • AA. Use Vertex Al Pipelines to execute the experiments. Query the results stored in MetadataStore using the Vertex Al API.
  • BB. Use Vertex Al Training to execute the experiments. Write the accuracy metrics to BigQuery, and query the results using the BigQuery API.
  • CC. Use Vertex Al Training to execute the experiments. Write the accuracy metrics to Cloud Monitoring, and query the results using the Monitoring API.
  • DD. Use Vertex Al Workbench user-managed notebooks to execute the experiments. Collect the results in a shared Google Sheets file, and query the results using the Google Sheets API.
  Show answer
• #158

  Your team successfully trained and tested a DNN regression model, but six months post-deployment, its performance has declined due to changes in the input data distribution. What approach should you take to tackle these differences in the input data in the production environment?

  • AA. Create alerts to monitor for skew, and retrain the model.
  • BB. Perform feature selection on the model, and retrain the model with fewer features.
  • CC. Retrain the model, and select an L2 regularization parameter with a hyperparameter tuning service.
  • DD. Perform feature selection on the model, and retrain the model on a monthly basis with fewer features.
  Show answer
• #159

  You are tasked with creating an ML model for a social media platform to determine whether a user's uploaded profile photo complies with the requirements. The objective is to provide users with feedback regarding the compliance of their pictures. What approach should you take in constructing the model to minimize the risk of incorrectly accepting a non-compliant image?

  • AA. Use Vertex AI AutoML to optimize the model’s recall to minimize false negatives.
  • BB. Use Vertex AI AutoML to optimize the model’s F1 score to balance the accuracy of false positives and false negatives.
  • CC. Use Vertex AI Workbench user-managed notebooks to build a custom model with three times as many examples of pictures that meet the profile photo requirements.
  • DD. Use Vertex AI Workbench user-managed notebooks to build a custom model with three times as many examples of pictures that do not meet the profile photo requirements.
  Show answer
• #160

  You are conducting experiments with a built-in distributed XGBoost model in Vertex AI Workbench user-managed notebooks. To split your data into training and validation sets, you use the following BigQuery queries: CREATE OR REPLACE TABLE ‘myproject.mydataset.training‘ AS (SELECT FROM ‘myproject.mydataset.mytable‘ WHERE RAND() <= 0.8); CREATE OR REPLACE TABLE ‘myproject.mydataset.validation‘ AS (SELECT FROM ‘myproject.mydataset.mytable‘ WHERE RAND() <= 0.2); After training the model, you achieve an area under the receiver operating characteristic curve (AUC ROC) value of 0.8. However, after deploying the model to production, you observe that the model's performance has dropped to an AUC ROC value of 0.65. What is the most likely problem occurring?

  • AA. There is training-serving skew in your production environment.
  • BB. There is not a sufficient amount of training data.
  • CC. The tables that you created to hold your training and validation records share some records, and you may not be using all the data in your initial table.
  • DD. The RAND() function generated a number that is less than 0.2 in both instances, so every record in the validation table will also be in the training table.
  Show answer

Want to track your score, take timed mock exams and get AI explanations? Create a free account