Table 5 Comparative results.
From: An integrating RAG-LLM and deep Q-network framework for intelligent fish control systems
System variant | Mean fish weight Gain (g) | Std. Dev. (g) | Decision error rate (%) |
|---|---|---|---|
Human Expert (Baseline) | 382.5 | 1.25 | – |
RAG-LLM only | 374.8 | 1.10 | 8.0 |
DQN only | 379.0 | 1.18 | 5.2 |
RAG-LLM + DQN (this work) | 389.3 | 1.20 | 1.8 |
