Case study
Project Qoconut
Generated QRNG datasets with Qiskit simulators and IBM hardware, then compared them with pseudorandom data through statistical tests and classification.
- Role:
- Quantum circuit and machine-learning contributor
- Timeline:
- SC Quantathon, Oct 2024
- Stack:
- Python, Qiskit, Jupyter, Machine Learning
How it works
Hadamard circuits generate bit strings through Aer simulation and IBM hardware. After normalization and a stratified split, a CNN and Random Forest classify quantum and pseudorandom samples; Logistic Regression combines their predictions.
Results and takeaways
- Running the same circuit in a simulator and on IBM hardware made device noise part of the experiment instead of an abstract caveat.
- Randomness tests measured sequence properties, while the classifier tested whether the two data sources contained learnable differences.