Training ML

Question 1

Custom Training: Can manually optimize model performance with hyperparameter tuning?

Answer 1

Yes. You can tune the model during each training run for experimentation and comparison.

Question 2

AutoML: Data science expertise needed?

Answer 2

No.

Question 3

AutoML: Time to trained model?

Answer 3

Lower. Less data preparation is required, and no development is needed.

Question 4

AutoML: Limits on machine learning objectives?

Answer 4

Yes, you must target one of AutoML’s predefined objectives.

Question 5

BigQueryML: Can manually optimize model performance with hyperparameter tuning?

Answer 5

Yes. BigQuery ML supports hyperparameter tuning when training ML models using CREATE MODEL statements.

Question 6

AutoML: Limits on data size?

Answer 6

Yes. AutoML uses managed datasets; data size limitations vary depending on the type of dataset.

Question 7

What are the possible solutions to working with a sparse dataset?

Answer 7

1. Use a model that supports training with sparse datasets (e.g. Wide&Deep, Autoencoder) 2. Remove features with low variance (e.g. use Lasso or L1 regularization) 3. Use a Dimensionality reduction method (e.g. PCA) to make sparse datasets dense, extracting principal components with the most variance.

Question 8

Custom Training: Limits on data size?

Answer 8

For unmanaged datasets, no. Managed datasets have the same limits as managed dataset objects created in and hosted by Vertex AI and are used to train AutoML models.

Question 9

Custom Training: Programming ability needed?

Answer 9

Yes, to develop the training application.

Question 10

BigQueryML: Data science expertise needed?

Answer 10

No.

Question 11

BigQueryML: Limits on data size?

Answer 11

Yes. BigQuery ML enforces appropriate quotas on a per-project basis.

Question 12

What are the challenges with sparse datasets?

Answer 12

Overfitting, High memory usage, Computational Complexity, Inaccurate results.

Question 13

BigQueryML: Programming ability needed?

Answer 13

SQL programming ability required to build, evaluate, and use the model in BigQuery ML.

Question 14

AutoML: Can manually optimize model performance with hyperparameter tuning?

Answer 14

No. AutoML does some automated hyperparameter tuning, but you can’t modify the values used.

Question 15

Custom Training: Can control aspects of the training environment?

Answer 15

Yes. You can specify aspects of the environment such as Compute Engine machine type, disk size, machine learning framework, and number of nodes.

Question 16

BigQueryML: Limits on machine learning objectives?

Answer 16

Yes.

Question 17

Custom Training: Limits on machine learning objectives?

Answer 17

No.

Question 18

BigQueryML: Time to trained model?

Answer 18

Lower. Model development speed is increased since you don’t need to build the infrastructure required for batch predictions or model training, as BigQuery ML leverages the BigQuery computational engine.

Question 19

AutoML: Programming ability needed?

Answer 19

No, AutoML is codeless.

Question 20

Custom Training: Data science expertise needed?

Answer 20

Yes, to develop the training application and also to do some of the data preparation like feature engineering.

Question 21

What is an example of a sparse dataset?

Answer 21

Netflix recommendation system. One-hot encoded categorical dataset for ‘Have you watched this movie before?’ out of 100 movies.

Think why this is sparse.

Question 22

BigQueryML: Can control aspects of the training environment?

Answer 22

No.

Question 23

AutoML: Can control aspects of the training environment?

Answer 23

Limited. For image and tabular datasets, you can specify the number of node hours to train for, and whether to allow early stopping of training.

Question 24

Custom Training: Time to trained model?

Answer 24

Higher. More data preparation is required, and training application development is needed.