[PMLE CERTIFICATE - EXAMTOPIC] DUMPS Q29-Q32

EXAMTOPIC DUMPS Q29-Q32; online prediction architecture on AI platform, Monitoring Data value Skews results in Bad performance in production

Q 29.

You have trained a model on a dataset that required computationally expensive preprocessing operations. You need to execute the same preprocessing at prediction time. You deployed the model on AI Platform for high-throughput online prediction. Which architecture should you use?

Online prediction architecture on AI platform

• A. Validate the accuracy of the model that you trained on preprocessed data. Create a new model that uses the raw data and is available in real time. Deploy the new model onto AI Platform for online prediction.
• ⭕ B. Send incoming prediction requests to a Pub/Sub topic. Transform the incoming data using a Dataflow job. Submit a prediction request to AI Platform using the transformed data. Write the predictions to an outbound Pub/Sub queue.
• C. Stream incoming prediction request data into Cloud Spanner. Create a view to abstract your preprocessing logic. Query the view every second for new records. Submit a prediction request to AI Platform using the transformed data. Write the predictions to an outbound Pub/Sub queue.
• D. Send incoming prediction requests to a Pub/Sub topic. Set up a Cloud Function that is triggered when messages are published to the Pub/Sub topic. Implement your preprocessing logic in the Cloud Function. Submit a prediction request to AI Platform using the transformed data. Write the predictions to an outbound Pub/Sub queue.

Can Implement computationally expensive preprocessing operations in Apache Beam, and run them at scale using Dataflow

Dataflow
- a fully managed autoscaling service for batch and stream data processing.
- perform instance-level transformations, stateful full-pass transformations, and window-aggregation feature transformations.

• Most of the time where you need to execute a full transformation pipeline & a comparison between dataflow and cloud function : recommended to go with dataflow. It's a solution more prepared to solve those cases.\

Q 30.

Your team trained and tested a DNN regression model with good results. Six months after deployment, the model is performing poorly due to a change in the distribution of the input data. How should you address the input differences in production?

Monitoring Data value Skews results in Bad performance in production

• ⭕ A. Create alerts to monitor for skew, and retrain the model.
• B. Perform feature selection on the model, and retrain the model with fewer features.
• 🚩 C. Retrain the model, and select an L2 regularization parameter with a hyperparameter tuning service.
  → L2 regularization prevent overfitting which can potential maintain model performance if data distribution is little skewed. HOWEVER THE Q SAID TEST RESULT WAS GOOD.
• D. Perform feature selection on the model, and retrain the model on a monthly basis with fewer features.

Need to trigger a retraining of the model to capture Data value skews

• Data values skews: These skews are significant changes in the statistical properties of data, which means that data patterns are changing, and you need to trigger a retraining of the model to capture
  these changes.

Q 31.

You need to train a computer vision model that predicts the type of government ID present in a given image using a GPU-powered virtual machine on Compute Engine. You use the following parameters:

• Optimizer: SGD
• Image shape = 224ֳ-224
• Batch size = 64
• Epochs = 10
• Verbose =2
  During training you encounter the following error: ResourceExhaustedError: Out Of Memory (OOM) when allocating tensor. What should you do?

NEED TO REDUCE MEMORY USE for ML training with image data

Image Classification using GPU-powered virtual machine on Compute Engine, to solve the error ResourceExhaustedError: Out Of Memory (OOM) ⇒ NEED TO REDUCE MEMORY USE

• ❌ A. Change the optimizer.
  → Learning rate and optimizer shouldn't really impact memory utilisation.
• ⭕ B. Reduce the batch size.
• ❌ C. Change the learning rate.
• 🚩 D. Reduce the image shape.
  → Decreasing image size would work, but might be costly in terms final performance

Batch size, Image shape

Batch size
✔ Defines the number of data samples used to compute each update to the model's trainable parameters (i.e. weights and biases) 파라미터 업데이트 연산에 사용되는 데이터 표본의 개수를 의미한다.
✔ critical impact on training time and the resulting accuracy of the trained model 훈련시간과 모델의 정확도에 중요한 영향을 미친다.
✔ The larger the batch, the more samples propagate through the model in the forward pass. Since a batch size increase will require more GPU memory, a lack of GPU memory can prevent you from increasing the batch size. 배치사이즈가 클수록 훈련 단계별 연산량이 많아지는데, 배치 사이즈 조정을 하는 경우 GPU 메모리에 제한을 받을 수 있다.

• Overview of hyperparameter tuning
• train_batch_size in hyperparameter examples
• TensorFlow guide to mixed precision training
• Discussions

1. OOM 메모리가 부족 에러가 발생했으므로, 더 적은 메모리를 사용해 훈련하려면 reduce batch size, reduce image shape 조정을 시도해볼 수 있다.
2. AI Platform Training > Hyperparameters of built-in image classification
   • train_batch_size : # of images used in one training step. If this number is too big, the job may fail with out-of-memory (OOM). (Default: 32)
   • image_size : The image size (width and height) used for training. Note that the training job may be OOM if its value is too big. (Default: 224)
3. 모델 훈련에 사용한 하이퍼파라미터
   • Image shape = 224ֳ-224 디폴트 224 와 일치하지만,
   • Batch size = 64 디폴트값 32 보다 높으므로 배치 사이즈를 줄여 다시 훈련해볼 수 있다.

Q 32.

You developed an ML model with AI Platform, and you want to move it to production. You serve a few thousand queries per second and are experiencing latency issues. Incoming requests are served by a load balancer that distributes them across multiple Kubeflow CPU-only pods running on Google Kubernetes Engine (GKE). Your goal is to improve the serving latency without changing the underlying infrastructure. What should you do?

• ❌ A. Significantly increase the max_batch_size TensorFlow Serving parameter.
  → bigger batch size ; increase latency
• B. Switch to the tensorflow-model-server-universal version of TensorFlow Serving.
• ❌ C. Significantly increase the max_enqueued_batches TensorFlow Serving parameter.
  → bigger batch size ; increase latency
• D. Recompile TensorFlow Serving using the source to support CPU-specific optimizations. Instruct GKE to choose an appropriate baseline minimum CPU platform for serving nodes.
  → focusing on server performance which development env is higher than production env.