OBSERVATORY / TECHNOLOGY
Machine Learning
Machine learning develops models whose behavior is fitted from data rather than specified only through fixed rules.
How it works
Training procedures adjust model parameters to reduce an objective on examples, simulations, demonstrations or interaction data.
Why it matters
Learning can help machines perceive complex scenes, predict outcomes and adapt policies to varied inputs.
Advantages
- handles high-dimensional data
- can improve with representative data
- supports perception and control
Limitations
- data and validation requirements
- distribution shift
- outputs can be difficult to interpret
Connected records
- RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control
- Open X-Embodiment: Robotic Learning Datasets and RT-X Models
- Diffusion Policy: Visuomotor Policy Learning via Action Diffusion
- Learning Hand-Eye Coordination for Robotic Grasping with Deep Learning and Large-Scale Data Collection
Source ledger
| Publisher and record | Scope | Evidence | Verified |
|---|---|---|---|
| NIST: Artificial Intelligence ↗ | general | Confirmed · Primary | Aug 20, 2026 |