Artificial Intelligence By Example

Training the CRLMM

In Chapter 9, Getting Your Neurons to Work, and Chapter 11, Conceptual Representation Learning, the CRLMM program CNN_STRATEGY_MODEL.py was trained to identify gaps in outputs on the conveyor belt of a food processing factory. Chapter 11, Conceptual Representation Learning, introduced CRLMMs using Γ (gap concept) to illustrate one.

In the previous, chapters, conceptual subsets around Γ (gap concepts) were designed as follows:

• Γ={g1,g2,g3,g4,g5,g6} which contains pg_i (p=positive) and ng_i(n=negative subsets)
• ng1 is a subset of Γ; ng1={missing, not enough, slowing production down...bad},
• pg2 is a subset of Γ;pg2=pg2={good production flow, no gap}
• g2={loaded}
• g3={unloaded},
• pg4={traffic jam, heavy traffic...too much traffic} ...