QPoland Global Quantum Hackathon 2025 Project

Project Title: Comparing Classical and Quantum-Inspired Kernels for Molecular Classification
Team: The Cats Cradle
Track: QML
Date: October 18, 2025 - October 26, 2025
Website: https://www.qaif.org/contests/qpoland-global-quantum-hackathon

Overview

This repository presents a comparative study of classical graph kernels and a quantum-inspired Trotterized quantum walk (QW) embedding for molecular and protein graph classification. The project leverages datasets such as AIDS, PROTEINS, NCI1, PTC-MR, and MUTAG, utilizing the QURI Parts framework to simulate quantum-inspired features. A key innovation is the ego-graph decomposition technique, which reduces qubit requirements while preserving functional-group topology, enhancing classification accuracy.

Deliverables

• Source Code: Repository src folder
• Report (GitHub Pages): https://jajapuramshivasai.github.io/QML-Molecular-Classification/
• Report (PDF in repo): QML-Molecular-Classification_report.pdf
• Video: https://drive.google.com/file/d/1p1539dZOyex4KxVeZW4sIBkU02wZt9MJ/view?usp=sharing

Quick Start

1. Clone the repository and install dependencies:

   git clone https://github.com/jajapuramshivasai/QML-Molecular-Classification.git
   cd QML-Molecular-Classification
   pip install -r requirements.txt

2. Run experiments:
   • Jupyter Notebook: jupyter notebook
   • Main Notebook: notebooks/sub_main.ipynb
   • Demo: notebooks/quantum_embeddings_demo.ipynb

Contents

• notebooks/: Experiment scripts (e.g., sub_main.ipynb)
• src/: Code for kernels, embeddings, and utilities
• pages/index.html: Project report for GitHub Pages
• experiment_results.json: Results and logs (if available)

Datasets

• TU benchmarks (AIDS, PROTEINS, NCI1, PTC-MR, MUTAG) via PyTorch Geometric TUDataset
• Preprocessing scripts in notebooks/ and src/

Implemented Methods

• Ego-QW: QURI-inspired ego-graph quantum walk embeddings
• CTQW: Continuous-time quantum walk spectral features
• Classical Baselines: Weisfeiler-Lehman subtree kernel, Shortest-Path kernel
• Classifier: SVM with RBF kernel, nested cross-validation for hyperparameters

Evaluation

• Nested 5-fold cross-validation
• Metrics: Mean ± std of Accuracy and F1-score
• Features standardized, class weights balanced
• Reproducible with fixed seeds (Python, NumPy, PyTorch)

Hardware/Software/Simulator Platforms

• We run our main notebook on Colab 12.7gb, CPU
• Simulator: QURI Parts to run our Quantum routines. We did not use real Quantum hardware.

Team

• Jajapuram Shiva Sai (@frosty)
• Amon Koike (@thedaemon_AK)
• Ramesh Makwana (@Ramesh Makwana)
• Dr. Sushant Tapase (@Dr-Sushant)

Contribution Guide

Notebooks

• Edit in Google Colab: Open notebooks/sub_main.ipynb, modify, and save/export to the repo.
• Workflow:
  1. Open in Colab via the notebook link.
  2. Edit and run end-to-end.
  3. Commit updated .ipynb with a change description.

Website (GitHub Pages)

• Edit pages/index.html.
• Preview locally:

  python3 -m http.server 8000 --directory Docs
  open http://localhost:8000/index.html

• Push changes to the default branch for updates.

Contact

Raise issues or PRs here, or contact team members via Discord handles.

References

1. X. Ai et al., “Towards Quantum Graph Neural Networks: An Ego‑Graph Learning Approach,” arXiv:2201.05158v3 (2024). HTML
2. D. Aharonov, A. Ambainis, J. Kempe, U. Vazirani, “Quantum Walks on Graphs,” STOC’01; arXiv:quant‑ph/0012090. PDF
3. C. Kluber, “Trotterization in Quantum Theory,” arXiv:2310.13296. PDF
4. N. Shervashidze et al., “Weisfeiler‑Lehman Graph Kernels,” JMLR 12, 2539–2561 (2011). PDF
5. K. M. Borgwardt, H.‑P. Kriegel, “Shortest‑path kernels on graphs,” ICDM 2005. PDF
6. PyTorch Geometric TUDataset. docs
7. PyTorch Geometric utils (to_networkx, conversions). utils
8. QURI Parts simulator and states. simulator, states
9. Continuous‑time quantum walk overview. overview
10. scikit‑learn SVC and RBF kernel. SVC, rbf_kernel