AI-900 Practice Question: Describe fundamental principles of machine learning on Azure

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  A system that automatically retrains models on a fixed daily schedule

  Why it's wrong here

  Fixed daily retraining is a form of MLOps lifecycle automation and operational cadence, not AutoML itself. AutoML is specifically concerned with automating the search over model families, preprocessing pipelines, and hyperparameter configurations to optimize a given metric. While a schedule can trigger an AutoML run, the scheduling mechanism alone neither selects algorithms nor tunes hyperparameters, so it misses the core definitional feature of AutoML.

• ✓

  Automatically iterating through algorithms and hyperparameters to find the best model for a dataset

  Why this is correct

  AutoML, or Automated Machine Learning, systematically explores a defined search space of candidate algorithms (e.g., linear regression, tree ensembles, deep networks) and their hyperparameter configurations. It uses techniques like Bayesian optimization, random search, or early termination to evaluate many candidates against a validation metric and automatically selects the best-performing model. This directly replaces the manual, iterative trial-and-error performed by a data scientist, which is the central capability that defines AutoML.

• ✗

  Automatically labelling training data using existing model predictions

  Why it's wrong here

  Automatically labeling training data using existing model predictions describes 'pseudo-labeling' or ML-assisted data labeling, which is a semi-supervised learning or human-in-the-loop annotation technique. AutoML does not create labels; it consumes labeled training data to search over models and hyperparameters. Labeling deals with data preparation and ground truth generation, whereas AutoML addresses the model selection and optimization phase, so this option conflates two separate stages of the ML pipeline.

• ✗

  A robot that physically connects GPU hardware for distributed training

  Why it's wrong here

  A robot physically connecting GPU hardware for distributed training is a hardware provisioning and infrastructure orchestration task, not an AutoML capability. AutoML operates purely at the software level, exploring algorithms and hyperparameters within a given computational environment. While AutoML runs may exploit distributed GPU clusters to speed up the search, the automation of physical hardware assembly is unrelated to the intelligent model-selection logic that AutoML provides, making this option definitively incorrect.

Go deeper

Related to this question