Gradient Descent for Reverse Fisher Estimation
Descriptions may be stale. Consult the linked Lean sources.
Gradient Descent for Reverse Fisher Estimation
We consider estimating a hidden vector
Equip
Our assumptions:
Assumption 1 (Signal bounds). The fixed target
belongs to the prescribed Euclidean annulus:
Our result is algorithm-specific: we analyze gradient descent. The
algorithm chooses its initialization
The required accuracy concerns recovery up to sign. The output must
be measurable and satisfy
Results
In
| Result | Oracle complexity | Novelty | Limitations |
|---|---|---|---|
| (1) SymmetricReverseFisherGlobalGD (Ours) (review) | |||
| (2) Derived bound: adaptive GD | (1): No established improvement. | (1): Larger budget; needs
|
|
| (3) GD (Tyurin 2026),
Thm. 3.2/E.1; two untied students, teachers |
(1): Scope of
constant-step analysis extends to zero separation. For |
(1): Only tied, antipodal components; one fixed terminal iterate, whereas (3) controls every subsequent iterate and permits general component counts. | |
| (4) GD (Tyurin 2026),
Thm. 2.1/Cor. 2.5; special case |
(1): Scope extends from a
single Gaussian teacher to every |
(1): At |
Abbr: