Table 1 Comparative summary of related works.
From: An integrating RAG-LLM and deep Q-network framework for intelligent fish control systems
Study | Methodology | Task domain | Actuation / Automation | Explainability | Adaptivity (RL) | LLM |
|---|---|---|---|---|---|---|
Chahid et al.15 | Q-learning | Feed scheduling, growth control | Manual response | Low | Yes | No |
Chen et al.14 | DQN | Fish school movement modeling | Simulation only | Low | Yes | No |
Metin et al.19 | Temporal Fusion Transformer (TFT) | Forecasting nitrate levels | No control layer | Moderate | No | No |
Kim et al.7 | AI-RCAS + vision | Species recognition, catch check | Diagnostic only | High | No | No |
This work | RAG-LLM + DQN ensemble | Full aquaculture automation | Full actuator control | Moderate–High | Yes | Yes |
