Cognition · Software component
Episodic Anomaly Scorer
Software componentCognitionCognition & Memoryarc:EpisodicAnomalyScorer
A cognition component that scores a new event's deviation from an entity's personal baseline by comparing it with the closest retrieved normal-pattern episodes across amount, time, counterparty and frequency signals.
Responsibility. Computes a personalized anomaly score for a new event against episodic history.
Also known as: Personalized baseline anomaly detector, Fraud anomaly scorer
Relationships
invokes dependency
writes dependency
escalates to dynamic
Design guidance
- SHOULD compare events only against the same entity's episodic history rather than global averages.
- SHOULD store confirmed legitimate outliers as marked episodes to reduce repeat false positives.
Quantitative guidance
As stated by the sources; verify before use.
- Example: $15K wire at 2 AM to a new foreign recipient was 3x the customer's prior max transfer (Ch5.7).
Classification
- Patterns
- Personalized baselines from episodic accumulationEmbedding-distance anomaly scoringLearning from confirmed false positives
- Quality attributes
- Functional suitability: correctness and validity (ISO/IEC 25010 | NIST AI RMF: valid)Interaction capability (ISO/IEC 25010)
- Risks mitigated
- False positives from universal rulesMissed fraud that global rules consider normal
Sources
- Ch5.7: T. Nguyen, "Episodic Memory," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 5.7. ISBN: 9798244538229.