Skip to content

Model Training & Adaptation

Deep reinforcement learning

AI learning through rewards and penalties using a large network of connected calculations.

Example

An AI learns actions in a practice environment by earning rewards.

Why people use it

It lets AI learn complex action choices through repeated attempts and feedback.

What you'll hear

“The agent improves by trying things and earning rewards.”

What this means for you

Check its behavior in the conditions where it will actually be used.

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

Can it learn from failed attempts?
Yes. Feedback about an unsuccessful action can help it choose differently later, if that feedback is useful.
Does success in a simulation guarantee success outside it?
No. Real conditions can differ from the simplified environment used for practice.
Can the reward encourage an unwanted shortcut?
Yes. The system may learn an easy way to earn points that misses the intended goal.

Related terms

Still have questions?

Up to 500 characters.

Ask LATHIC about AI. Relevant glossary entries may be included.

Your question, the glossary entries it matches, and a rotating pseudonymous identifier go to Microsoft Azure’s OpenAI service through Vercel AI Gateway to generate an answer. Zero retention and no training are required of the provider, and LATHIC does not save your question or answer. Privacy Notice