Learning Feedback — Conceptual Overview

Audience: Implementers, ML Engineers Governance Rule: DGP-30

1. What Learning Feedback Refers To

Learning Feedback in MPLP refers to the improvement dimension that concerns how agent experiences are captured for reinforcement learning and fine-tuning.

Learning Feedback is not a training pipeline. It is a conceptual area for structured experience capture.

2. Conceptual Areas Covered by Learning Feedback

Conceptual Area	Description
Learning Samples	Relates to structured experience records
User Feedback	Concerns approval/rejection signals from Confirm
Intent Resolution	Is involved in intent-to-plan mapping samples
Delta Impact	Relates to change prediction samples

3. What Learning Feedback Does NOT Do

❌ Define training algorithms
❌ Mandate specific ML frameworks
❌ Prescribe model architectures
❌ Define reward functions

4. Where Normative Semantics Are Defined

Normative Source	What It Covers
Learning Schemas (`learning/*.schema.json`)	Sample structures
Learning Invariants	12 rules for sample structure
Confirm Module	User feedback capture

5. Conceptual Relationships

6. Reading Path

Learning Overview

Note: The Learning Sample Specification is currently under revision.

Governance Rule: DGP-30 See Also: Learning Feedback Anchor (Normative)

1. What Learning Feedback Refers To​

2. Conceptual Areas Covered by Learning Feedback​

3. What Learning Feedback Does NOT Do​

4. Where Normative Semantics Are Defined​

5. Conceptual Relationships​

6. Reading Path​