Model Training & Adaptation
Automated machine learning
Tools that handle some of the work of building and choosing AI prediction systems.
Example
A service tries several AI system types and training settings on a collection of information.
Why people use it
It can speed up the trial and comparison work involved in building AI.
What you'll hear
“Let the software try several approaches first.”
What this means for you
Review the chosen AI system against the actual use case and error costs.
Can you control it?
Developer-only
The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.
Common questions
- Does AutoML remove the need for problem definition and checking?
- No. People still choose appropriate data, goals and acceptance criteria.
- Does it choose the business goal for us?
- No. People still need to decide what success means and which mistakes matter.
- Can it pick something hard to explain?
- Yes. The best-scoring choice may not be the easiest for people to understand or maintain.