AI Foundations
Deep neural network
An AI network that processes information through multiple connected stages of learned calculations.
Example
AI examines a picture through several learned stages before labeling it.
Why people use it
It lets AI build complex patterns through several stages of learning.
What you'll hear
“How many stages does it use to process the information?”
What this means for you
Judge useful results rather than treating more stages as automatically better.
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 deeper automatically mean better?
- No. Data, design, training and task requirements determine performance.
- Does each layer have an obvious human meaning?
- Not necessarily. A useful internal pattern may be difficult to describe in everyday terms.
- Can training require substantial computing power?
- Yes. The amount depends on the network, the examples and how the learning is organized.