Hybrid Quantum-Classical Classifier on make_moons

A minimal, reproducible proof-of-concept hybrid quantum neural network (HQNN) for binary classification on the non-linear make_moons dataset.

Combines classical pre-processing with a variational quantum circuit (RealAmplitudes ansatz via Qiskit EstimatorQNN) wrapped as a PyTorch layer using TorchConnector.

Achieves 100.00% test accuracy after 50 epochs — demonstrating effective learning of the decision boundary in this toy setting.

Why This Project?

• Demonstrates seamless integration of Qiskit quantum circuits into PyTorch training loops
• Shows how small variational quantum layers can contribute to classification tasks
• Serves as a baseline for exploring quantum advantages in optimization-heavy domains (e.g., FinOps anomaly detection, portfolio modeling, multicloud resource allocation)

Tech Stack

• Quantum: Qiskit 1.x (RealAmplitudes ansatz, StatevectorEstimator, EstimatorQNN, TorchConnector)
• Classical/ML: PyTorch (nn.Module, Adam optimizer), scikit-learn (make_moons, StandardScaler, train_test_split)
• Environment: Python 3.10+, Jupyter/Colab

Results

• Dataset: make_moons (n=100, noise=0.1)
• Train/test split: 80/20
• Model: Classical linear + tanh → 1-qubit RealAmplitudes (reps=3) → linear output
• Optimizer: Adam (lr=0.01)
• Epochs: 50
• Final test accuracy: 100.00% (perfect separation on this small, clean dataset)

How to Run

1. Open in Colab:
   

2. Install dependencies (first cell):

   !pip install qiskit qiskit-machine-learning qiskit-aer torch scikit-learn matplotlib --quiet