Skip to content

Latest commit

 

History

History
29 lines (16 loc) · 1.35 KB

File metadata and controls

29 lines (16 loc) · 1.35 KB

References & Future Directions

Future Directions

  1. Real-world validation — Test on actual clinical documents (with appropriate ethics approval)

  2. Automated fix application — Close the loop from proposed fixes to implementation

  3. Cross-domain transfer — Do discovered weaknesses generalize to non-clinical German text?

  4. Active learning — Use disagreement analysis to prioritize human annotation

References

Agentic AI Research

  • ReAct: Yao, S., et al. (2022). "ReAct: Synergizing Reasoning and Acting in Language Models." ICLR 2023. arXiv:2210.03629

  • Reflexion: Shinn, N., et al. (2023). "Reflexion: Language Agents with Verbal Reinforcement Learning." NeurIPS 2023. arXiv:2303.11366

  • Multi-Agent Systems: Han, J., et al. (2024). "LLM Multi-Agent Systems: Challenges and Open Problems." arXiv:2402.03578

  • Curriculum Learning: Bengio, Y., et al. (2009). "Curriculum Learning." ICML 2009.

Tools & Frameworks