Machine Learning

Question 1

TensorFlow

Answer 1

Deep learning OS deployed on Compute Engine - C++ APIs use GPUs.

Use case:
Complex, highly customised ML models.

Question 2

ML Engine

Answer 2

Multi platform, NoOps ML tool - can implement tools such as TensorFlow, sci kit-learn & XGboost.

Use cases:
Low latency online machine learning predictions
High latency, high accuracy batch predictions

Question 3

ML APIs

Answer 3

Simple APIs for common ML tasks - Speech, Vision & Language

Question 4

AutoML

Answer 4

No ops, GUI based ML adjustment or creation.

Use cases:
Training existing ML models for new use (e.g adding a keyword to a speech model) or creating simple ML models. But