This repository contains the code, presentation, and research paper for our project on Conformal Prediction for Signal Temporal Logic (STL) Runtime Verification, completed as part of the CS637 - Embedded and Cyber-Physical Systems course.
Runtime verification is a lightweight formal method used to verify system behaviors during execution. In this project, we explore the use of Conformal Prediction to enhance the robustness of STL-based runtime verification. Conformal prediction provides a statistically valid way to associate confidence levels with predictions, offering a significant advantage in the context of runtime safety-critical systems.
- Paper: Presented the theoretical framework and case studies illustrating the application of conformal prediction for STL runtime verification.
- Implementation: Recreated the code from the paper, ensuring reproducibility and better understanding of the methods.
- Presentation: Summarized the paper's contributions, challenges, and our implementation results during the course.
📁 code/ # Implementation of the algorithms and case studies
📁 presentation/ # Slides used during the course presentation
📁 paper/ # Original research paper README.md # Project overview and instructions
- Clone the repository:
git clone https://github.com/praneatdata/conformal-prediction-STL.git
- Navigate to the code directory:
cd conformal-prediction-STL\code\example_5_online_verification\CP Online Verification Example 5.ipynb
- Run the Jupyter Notebook:
jupyter notebook
- Follow the instructions in the notebook to understand the implementation and results.
The implementation reproduces the results from the paper and validates the applicability of conformal prediction in runtime verification scenarios.
We extend our gratitude to the CS637 course instructor, Prof. Indranil Saha for his guidance and feedback during the project.