RMSprop
Algorithms AIMech:RMSprop DRAFT
An adaptive gradient optimization method that normalizes coordinate updates using an exponentially weighted estimate of squared gradients.
Identity Notes & Concept Scope
RMSprop decays older squared-gradient information, unlike cumulative Adagrad. Centered and momentum settings are variants within this scoped record.
Claims & Grounded Evidence
1 Claim
Mechanism
#update-rule
An exponential moving average of squared gradients supplies the update denominator; the documented centered option subtracts a squared mean-gradient estimate, and momentum is optional.
⚠️ Curated Limitations & Negative Results
- The cited implementation adds epsilon after the square root and documents a different ordering in TensorFlow.
- No universal improvement or convergence guarantee is inferred from implementation documentation.
Artificial Intelligence Ontology (AIO) Grounding
Mapped AIO Classes
| AIO Target Class | Mapping Relation | Mapping Rationale |
|---|---|---|
AIO:TrainingStrategy
TrainingStrategy
|
related_match | This algorithm can implement parameter fitting, but the pinned TrainingStrategy definition uses a preprocessing genus. This is a contextual training relationship, not equivalence or a subclass assertion. |
Evidence Sources & Primary Literature
1 Source| Source ID | Publication / Document | Type / Access | Reference & Checksum |
|---|---|---|---|
pytorch-docs
|
PyTorch contributors
|
official documentation
documentation
|
sha256:b3aeb10dd57d…
|
Curation Provenance
Curator: Codex
Method: ai_assisted
Model: GPT-6
Created: 2026-10-11T03:25:18Z
Updated: 2026-10-11T03:25:18Z
Source File:
data/curated/RMSprop.yaml