Quantum Support Vector Machine (QSVM)

A comparative study of classical and quantum Support Vector Machines implemented with PennyLane and scikit-learn.

📋 Overview

This repository explores the application of quantum computing to machine learning through Support Vector Machines. It compares classical SVM kernels (linear, polynomial, RBF, sigmoid) with quantum kernels built using parameterized quantum circuits.

🗂️ Repository Structure

.
├── svm.ipynb          
├── qsvm.ipynb       
├── requirements.txt   
├── .gitignore  
└── README.md

svm.ipynb

• Classical SVM implementation
• Focus on Iris dataset
• Kernel comparison (linear, poly, RBF, sigmoid)
• Visualization of decision boundaries

qsvm.ipynb

• Quantum kernel implementation using PennyLane
• Support for multiple datasets:
  • Iris: 3-class flower classification
  • Breast Cancer Wisconsin: Binary tumor classification
  • Banknote Authentication: Counterfeit detection
  • Glass Identification: Forensic glass type
  • Haberman Survival: Cancer survival prediction
• Side-by-side comparison of classical and quantum kernels
• Quantum circuit visualization
• Kernel matrix analysis

🎯 Key Features

• Quantum Feature Mapping: Encodes classical data into quantum states using angle embedding
• Quantum Kernel: Computes similarity as K(x₁, x₂) = |⟨0|U†(x₂)U(x₁)|0⟩|²
• PCA Integration: Dimensionality reduction for high-dimensional datasets
• Interactive Visualizations: Kernel matrices, PCA projections, confusion matrices
• Performance Metrics: Accuracy, confusion matrices, classification reports