Quantum Clustering project for QSOF 2024 Spring mentorship program
This project contrasts classical clustering techniques with quantum-inspired clustering using the Schrödinger equation and quantum clustering via the D-Wave quantum annealer to evaluate efficiency and accuracy on the Crab and Cancer datasets.
- Lower sigma values tend to create more localized clusters, which can capture finer details.
- Higher sigma values result in broader clusters, which are more robust to noise but may overlook finer details.
- Scanning over sigma values provides insight in the appropriate choice of sigma value, where the number of clusters is stable.
- The Conjugate-Gradient and Broyden-Fletcher-Goldfarb-Shanno algorithms show reliable performance.
- Various optimization algorithms result in consistent trend in the number of clusters as a function of sigma values, increasing the confidence in the choice of sigma value.