Intrinsic dimensionality
The minimum number of dimensions needed to accurately represent a dataset's structure; simple tasks like 2D navigation have low intrinsic dimensionality, while complex tasks like 3D rotation have high intrinsic dimensionality
As defined in LeWorldModel: Stable End-to-End JEPA World Models From Pixels
Where it comes up · 1
Most important claims · 1 of 1Search all claims for “Intrinsic dimensionality” →
- The fundamental trade-off of forcing AI representations to match a rigid isotropic Gaussian distribution is that the mandatory geometric complexity causes LayWM to fail on simple tasks with low intrinsic dimensionality, such as a 2D dot navigating through a door (two-room environment).LeWorldModel: Stable End-to-End JEPA World Models From Pixels