PRIMUS Overview
Introduction to the Pattern Recognition and Integration Multi-Unit System
PRIMUS Overview
PRIMUS (Pattern Recognition and Integration Multi-Unit System) is a proposed architectural framework for AGI that focuses on the hierarchical integration of sensory information and cognitive patterns.
Concept and Inspiration
PRIMUS is inspired by the hierarchical organization of the biological brain, particularly the neocortex. It posits that intelligence arises from the interaction of many small, similar processing units arranged in a hierarchy.
Core Pillars
1. Pattern Recognition Units (PRUs)
The fundamental building blocks of PRIMUS. Each unit is responsible for identifying specific patterns in its input and passing its "confidence" to higher levels.
2. Hierarchical Integration
Patterns found at lower levels (e.g., edges, colors in a vision unit) are integrated at higher levels to form complex concepts (e.g., faces, objects, scenes).
3. Multi-Unit System
Unlike a single monolithic neural network, PRIMUS is designed as a system of many semi-autonomous units that can be added, updated, or removed dynamically.
Relationship with Hyperon
While Hyperon provides the "knowledge storage and reasoning" layer (AtomSpace + MeTTa), PRIMUS can be viewed as a specific way to organize the perceptual and active side of an AGI system. PRIMUS units could be implemented as specialized MeTTa scripts or as neural networks integrated into the AtomSpace.
Goals of PRIMUS
- Robustness: The system should continue to function even if some units fail.
- Scalability: New units can be added to handle new sensory modalities or tasks.
- Real-Time Operation: Hierarchical processing allows for fast, reactive behavior at lower levels while slower, deliberative reasoning happens at higher levels.