Google Professional Machine Learning Engineer

• #181

  You are tasked with a dataset that encompasses customer transactions, and your objective is to construct an ML model for forecasting customer purchase patterns. Your plan involves creating the model within BigQuery ML and subsequently exporting it to Cloud Storage for online prediction. Upon reviewing the data, you observe the presence of categorical features such as product category and payment method. Your priority is to deploy the model swiftly. What steps should you take to achieve this goal?

  • AA. Use the TRANSFORM clause with the ML.ONE_HOT_ENCODER function on the categorical features at model creation and select the categorical and non-categorical features.
  • BB. Use the ML.ONE_HOT_ENCODER function on the categorical features and select the encoded categorical features and non-categorical features as inputs to create your model.
  • CC. Use the CREATE MODEL statement and select the categorical and non-categorical features.
  • DD. Use the ML.MULTI_HOT_ENCODER function on the categorical features, and select the encoded categorical features and non-categorical features as inputs to create your model.
  Tampilkan jawaban
• #182

  As an ML engineer at a bank responsible for developing an ML-based biometric authentication system for the mobile application, you've been tasked with verifying a customer's identity based on their fingerprint. Fingerprints are considered highly sensitive personal information and cannot be downloaded and stored in the bank's databases. What machine learning strategy should you suggest for training and deploying this ML model?

  • AA. Data Loss Prevention API
  • BB. Federated learning
  • CC. MD5 to encrypt data
  • DD. Differential privacy
  Tampilkan jawaban
• #183

  You need to train an XGBoost model on a small dataset and your training code has custom dependencies. To minimize the startup time of your training job, how should you configure your Vertex AI custom training job?

  • AA. Store the data in a Cloud Storage bucket and create a custom container with your training application, which reads the data from Cloud Storage to train the model.
  • BB. Use the XGBoost prebuilt custom container, create a Python source distribution that includes the data and installs dependencies at runtime, and train the model by loading the data into a pandas DataFrame.
  • CC. Create a custom container that includes the data and train the model by loading the data into a pandas DataFrame.
  • DD. Store the data in a Cloud Storage bucket, use the XGBoost prebuilt custom container for your training application, create a Python source distribution that installs the dependencies at runtime, and read the data from Cloud Storage to train the model.
  Tampilkan jawaban
• #184

  You are employed by a startup that manages various data science workloads. Currently, your compute infrastructure is on-premises, and the data science workloads rely on PySpark. Your team is planning to migrate these data science workloads to Google Cloud. To initiate a proof of concept for migrating one data science job to Google Cloud while minimizing cost and effort, what should be your initial step?

  • AA. Create a n2-standard-4 VM instance and install Java, Scala, and Apache Spark dependencies on it.
  • BB. Create a Google Kubernetes Engine cluster with a basic node pool configuration, install Java, Scala, and Apache Spark dependencies on it.
  • CC. Create a Standard (1 master, 3 workers) Dataproc cluster, and run a Vertex AI Workbench notebook instance on it.
  • DD. Create a Vertex AI Workbench notebook with instance type n2-standard-4.
  Tampilkan jawaban
• #185

  You're part of a rapidly growing social media company, where your team builds TensorFlow recommender models on an on-premises CPU cluster. With billions of historical user events and 100,000 categorical features in the data, you've observed increasing model training times as the data grows. Now, you're planning to migrate the models to Google Cloud and seek the most scalable approach that minimizes training time. What should you do?

  • AA. Deploy the training jobs using TPU VMs with TPUv3 Pod slices, and utilize the TPUEmbedding API.
  • BB. Deploy the training jobs within an autoscaling Google Kubernetes Engine cluster with CPUs.
  • CC. Deploy a matrix factorization model training job using BigQuery ML.
  • DD. Deploy the training jobs using Compute Engine instances with A100 GPUs, and utilize the tf.nn.embedding_lookup API.
  Tampilkan jawaban
• #186

  As you develop models to classify customer support emails, you initially created TensorFlow Estimator models using small datasets on your local system. To enhance performance, you now plan to train these models with larger datasets. For a seamless transition of your models from on-premises to Google Cloud, with minimal code refactoring and infrastructure overhead, what approach should you take?

  • AA. Use AI Platform for distributed training.
  • BB. Create a cluster on Dataproc for training.
  • CC. Create a Managed Instance Group with autoscaling.
  • DD. Use Kubeflow Pipelines to train on a Google Kubernetes Engine cluster.
  Tampilkan jawaban
• #187

  You need to execute a batch prediction on 100 million records in a BigQuery table with a custom TensorFlow DNN regressor model, and then store the predicted results in a BigQuery table. You want to minimize the effort required to build this inference pipeline. What steps should you take to achieve this?

  • AA. Import the TensorFlow model with BigQuery ML, and run the ml.predict function.
  • BB. Use the TensorFlow BigQuery reader to load the data, and use the BigQuery API to write the results to BigQuery.
  • CC. Create a Dataflow pipeline to convert the data in BigQuery to TFRecords. Run a batch inference on Vertex AI Prediction, and write the results to BigQuery.
  • DD. Load the TensorFlow SavedModel in a Dataflow pipeline. Use the BigQuery I/O connector with a custom function to perform the inference within the pipeline, and write the results to BigQuery.
  Tampilkan jawaban
• #188

  You need to develop a custom TensorFlow model for online predictions with training data stored in BigQuery. You want to apply instance-level data transformations to the data consistently during both model training and serving. How should you configure the preprocessing routine?

  • AA. Create a BigQuery script to preprocess the data and save the result to another BigQuery table.
  • BB. Create a Vertex AI Pipelines pipeline to read the data from BigQuery and perform preprocessing using a custom preprocessing component.
  • CC. Develop a preprocessing function that reads and transforms the data from BigQuery. Create a Vertex AI custom prediction routine that calls the preprocessing function during serving.
  • DD. Implement an Apache Beam pipeline that reads data from BigQuery and preprocesses it using TensorFlow Transform and Dataflow.
  Tampilkan jawaban
• #189

  You are employed by a gaming company that oversees a popular online multiplayer game featuring 5-minute battles between teams of 6 players. With a continuous influx of new players daily, your task is to develop a real-time model for automatically assigning available players to teams. According to user research, battles are more enjoyable when players of similar skill levels are matched together. What key business metrics should you monitor to evaluate the performance of your model?

  • AA. Average time players wait before being assigned to a team
  • BB. Precision and recall of assigning players to teams based on their predicted versus actual ability
  • CC. User engagement as measured by the number of battles played daily per user
  • DD. Rate of return as measured by additional revenue generated minus the cost of developing a new model
  Tampilkan jawaban
• #190

  You are an ML engineer at a manufacturing company. You need to build a model that identifies defects in products based on images of the product taken at the end of the assembly line. You want your model to preprocess the images with lower computation to quickly extract features of defects in products. Which approach should you use to build the model?

  • AA. Reinforcement learning
  • BB. Recommender system
  • CC. Recurrent Neural Networks (RNN)
  • DD. Convolutional Neural Networks (CNN)
  Tampilkan jawaban
• #191

  You have recently deployed a model to a Vertex AI endpoint, and you are encountering frequent data drift. To address this, you have enabled request-response logging and established a Vertex AI Model Monitoring job. However, you've noticed that your model is receiving higher traffic than initially anticipated. Your goal is to reduce the cost of model monitoring while still maintaining the ability to promptly detect drift. What step should you take?

  • AA. Replace the monitoring job with a DataFlow pipeline that utilizes TensorFlow Data Validation (TFDV).
  • BB. Replace the monitoring job with a custom SQL script designed to calculate statistics on the features and predictions within BigQuery.
  • CC. Decrease the sample_rate parameter in the RandomSampleConfig of the monitoring job.
  • DD. Increase the monitor_interval parameter in the ScheduleConfig of the monitoring job.
  Tampilkan jawaban
• #192

  You work for a retail company, and your task is to develop a model for predicting whether a customer will make a purchase on a given day. Your team has processed the company's sales data and created a table with specific columns, including customer ID, product ID, date, days since the last purchase, average purchase frequency, and a binary class indicating whether the customer made a purchase on the date in question. Your objective is to interpret the results of your model for individual predictions. What is the recommended approach?

  • AA. Create a BigQuery table. Use BigQuery ML to construct a boosted tree classifier. Examine the partition rules within the trees to comprehend how each prediction is guided through the tree structure.
  • BB. Create a Vertex AI tabular dataset. Train an AutoML model for predicting customer purchases. Deploy the model to a Vertex AI endpoint and enable feature attributions. Utilize the "explain" method to acquire feature attribution values for each individual prediction.
  • CC. Create a BigQuery table. Employ BigQuery ML to develop a logistic regression classification model. Interpret the feature importance by examining the coefficients of the model, with higher coefficient values indicating greater importance.
  • DD. Create a Vertex AI tabular dataset. Train an AutoML model for predicting customer purchases. Deploy the model to a Vertex AI endpoint. For each prediction, activate L1 regularization to identify non-informative features.
  Tampilkan jawaban
• #193

  You designed a 5-billion-parameter language model in TensorFlow Keras that used autotuned tf.data to load the data in memory. You created a distributed training job in Vertex AI with tf.distribute.MirroredStrategy, and set the large_model_v100 machine for the primary instance. The training job fails with the following error: The replica 0 ran out of memory with a non-zero status of 9. You want to fix this error without vertically increasing the memory of the replicas. What should you do?

  • AA. Keep MirroredStrategy. Increase the number of attached V100 accelerators until the memory error is resolved.
  • BB. Switch to ParameterServerStrategy, and add a parameter server worker pool with large_model_v100 instance type.
  • CC. Switch to tf.distribute.MultiWorkerMirroredStrategy with Reduction Server. Increase the number of workers until the memory error is resolved.
  • DD. Switch to a custom distribution strategy that uses TF_CONFIG to equally split model layers between workers. Increase the number of workers until the memory error is resolved.
  Tampilkan jawaban
• #194

  As an ML engineer at a bank, you've created a binary classification model using Vertex AI AutoML Tables to determine whether a customer will make timely loan payments, which is critical for loan approval decisions. Now, the bank's risk department has requested an explanation for why the model rejected a specific customer's loan application. What steps should you take in response to this request?

  • AA. Use local feature importance from the predictions.
  • BB. Use the correlation with target values in the data summary page.
  • CC. Use the feature importance percentages in the model evaluation page.
  • DD. Vary features independently to identify the threshold per feature that changes the classification.
  Tampilkan jawaban
• #195

  You were tasked with investigating failures in a production line component using sensor readings. Upon receiving the dataset, you found that less than 1% of the readings are positive examples representing failure incidents. Despite multiple attempts to train classification models, none of them have converged. How should you address the issue of class imbalance?

  • AA. Utilize the class distribution to generate 10% positive examples.
  • BB. Implement a convolutional neural network with max pooling and softmax activation.
  • CC. Downsample the data with upweighting to create a sample with 10% positive examples.
  • DD. Remove negative examples until the numbers of positive and negative examples are equal.
  Tampilkan jawaban
• #196

  You work at a gaming startup and have several terabytes of structured data stored in Cloud Storage, including gameplay time, user metadata, and game metadata. You need to build a model to recommend new games to users with minimal coding. What should you do?

  • AA. Load the data into BigQuery and use BigQuery ML to train an Autoencoder model.
  • BB. Load the data into BigQuery and use BigQuery ML to train a matrix factorization model.
  • CC. Load the data into a Vertex AI Workbench notebook and use TensorFlow to train a two-tower model.
  • DD. Load the data into a Vertex AI Workbench notebook and use TensorFlow to train a matrix factorization model.
  Tampilkan jawaban
• #197

  You work for a large hotel chain and have been asked to assist the marketing team in gathering predictions for a targeted marketing strategy. You need to make predictions about user lifetime value (LTV) over the next 20 days so that marketing can be adjusted accordingly. The customer dataset is in BigQuery, and you are preparing the tabular data for training with AutoML Tables. This data has a time signal that is spread across multiple columns. How should you ensure that AutoML fits the best model to your data?

  • AA. Manually combine all columns that contain a time signal into an array. AIlow AutoML to interpret this array appropriately. Choose an automatic data split across the training, validation, and testing sets.
  • BB. Submit the data for training without performing any manual transformations. AIlow AutoML to handle the appropriate transformations. Choose an automatic data split across the training, validation, and testing sets.
  • CC. Submit the data for training without performing any manual transformations, and indicate an appropriate column as the Time column. AIlow AutoML to split your data based on the time signal provided, and reserve the more recent data for the validation and testing sets.
  • DD. Submit the data for training without performing any manual transformations. Use the columns that have a time signal to manually split your data. Ensure that the data in your validation set is from 30 days after the data in your training set and that the data in your testing sets from 30 days after your validation set.
  Tampilkan jawaban
• #198

  To analyze user activity data from your company's mobile applications using BigQuery for data analysis, transformation, and ML algorithm experimentation, you must establish real-time data ingestion into BigQuery. What steps should you take to achieve this?

  • AA. Configure Pub/Sub to stream the data into BigQuery.
  • BB. Run an Apache Spark streaming job on Dataproc to ingest the data into BigQuery.
  • CC. Run a Dataflow streaming job to ingest the data into BigQuery.
  • DD. Configure Pub/Sub and a Dataflow streaming job to ingest the data into BigQuery.
  Tampilkan jawaban
• #199

  You recently developed a wide and deep model in TensorFlow, and you generated training datasets using a SQL script in BigQuery for preprocessing raw data. You now need to create a training pipeline for weekly model retraining, which will generate daily recommendations. Your goal is to minimize model development and training time. How should you develop the training pipeline?

  • AA. Use the Kubeflow Pipelines SDK to implement the pipeline. Employ the BigQueryJobOp component to run the preprocessing script and the CustomTrainingJobOp component to launch a Vertex AI training job.
  • BB. Use the Kubeflow Pipelines SDK to implement the pipeline. Utilize the DataflowPythonJobOp component for data preprocessing and the CustomTrainingJobOp component to initiate a Vertex AI training job.
  • CC. Use the TensorFlow Extended SDK to implement the pipeline. Use the ExampleGen component with the BigQuery executor for data ingestion, the Transform component for data preprocessing, and the Trainer component to launch a Vertex AI training job.
  • DD. Use the TensorFlow Extended SDK to implement the pipeline. Integrate the preprocessing steps into the input_fn of the model. Use the ExampleGen component with the BigQuery executor for data ingestion and the Trainer component to initiate a Vertex AI training job.
  Tampilkan jawaban
• #200

  You are employed at a pharmaceutical company in Canada, and your team has developed a BigQuery ML model for predicting the monthly flu infection count in Canada. Weather data is updated weekly, while flu infection statistics are updated monthly. Your task is to establish a model retraining policy that minimizes expenses. What would you recommend?

  • AA. Download the weather and flu data on a weekly basis. Configure Cloud Scheduler to trigger a Vertex AI pipeline for weekly model retraining.
  • BB. Download the weather and flu data on a monthly basis. Configure Cloud Scheduler to trigger a Vertex AI pipeline for monthly model retraining.
  • CC. Download the weather and flu data weekly. Configure Cloud Scheduler to trigger a Vertex AI pipeline for model retraining every month.
  • DD. Download weather data weekly and flu data monthly. Deploy the model on a Vertex AI endpoint with feature drift monitoring and retrain the model if a monitoring alert is triggered.
  Tampilkan jawaban

Ingin melacak skor, mengikuti ujian simulasi berwaktu, dan mendapat penjelasan AI? Buat akun gratis