Natural Language Analysis TRIZ Cases
- Selection-Inference Neural Networks With Interpretable Reasoning Traces
Alternating selection and inference steps expose a causal natural-language reasoning trace for safer control decisions and fault diagnosis.
- Vehicle Voice Interface Using SLM-LLM Routing to Curb Hallucinations
Classifying in-car speech between SLM and LLM paths improves response accuracy and reduces inappropriate or hallucinated outputs.