Self-Evolving Programs: A Novel Approach Leveraging LLMs and Quine Programs

Self-Evolving Programs: A Novel Approach Leveraging LLMs and Quine Programs

1. Self-Evolving Programs: A Novel Approach Leveraging LLMs and Quine Programs Presenter: Dr. Saghiri and Dr. Wang A. M. Saghiri and Nan Wang, "Self-Evolving Programs: A Novel Approach Leveraging LLMs and Quine Programs," in Proceedings of the International Conference on Computational Intelligence and Machine Learning Systems (ICCIMS 2024), July 29-31, 2024, Royal Military College of Canada & Université du Québec en Outaouais, Canada.

2. Outline

3. Introduction Rapid Evolution in Software Systems: Modern software evolves quickly. Need for adaptive and efficient coding methods. Focus Self-evolving programs Large Language Models (LLMs) Quine Programs.

4. Key Concepts

5. The Proposed Framework Component-Based Design: Predictive Analysis Module (PAM): Uses LLMs to predict threats and inefficiencies. Self-Replicating Code Generator (SRCG): Generates and modifies core code based on Quine Programs. Adaptive Strategy Executor (ASE): Adjusts software behavior based on PAM and SRCG inputs. Monitoring and Evaluation Unit (MEU): Assesses performance and fine-tunes the system.

6. Application in Blockchain

7. Simulation Results

8. Key Observation

9. Discussion

10. Conclusion and Future Work

11. References