Explains the information-geometric inner update, the paper’s diagonal Fisher approximation, and tests required before claiming convergence or numerical stability.
Research documentation: this page interprets a cited research source and defines evidence requirements. It does not claim a released Teleodynamic AI implementation.
Information geometry
Natural gradients precondition an ordinary gradient with the inverse Fisher information metric so a step is interpreted relative to the parameter manifold rather than raw coordinates.
Reported approximation
DE11 uses a diagonal Fisher approximation. The paper acknowledges that ignoring parameter correlations can be suboptimal and lists richer approximations as future work.
Verification
- Compare ordinary and natural-gradient traces.
- Check Fisher positivity, floors, and overflow behavior.
- Record convergence criterion and uncertainty.
- Test correlated features where the diagonal approximation should be stressed.
Scope
This starter page defines the questions, boundaries, evidence, and failure modes that should be recorded before a capability is presented as supported.
Engineering considerations
- Identify the source, version, target environment, and owner.
- Separate observed values from estimates and externally reported values.
- Record trade-offs, unsupported cases, and fallback behavior.
- Link performance statements to a compatible benchmark methodology.
Verification questions
- What exact artifact, revision, backend, and environment were reviewed?
- Which assumptions could change the result?
- Which data should be retained so another engineer can reproduce the conclusion?