Core Concepts
The author proposes the Quantum-SMOTE method, utilizing quantum computing techniques to address class imbalance in machine learning datasets. By introducing hyperparameters like rotation angle and minority percentage, the approach offers greater control over synthetic data generation.
Abstract
The paper introduces the Quantum-SMOTE method, leveraging quantum computing for class imbalance in machine learning datasets. It explores SMOTE variations and their impact on classification models. The study includes detailed explanations of swap tests, rotations, and synthetic data creation using quantum principles.
Unbalanced classification is a common issue in machine learning due to uneven class representation. Techniques like SMOTE aim to address this problem by generating artificial samples from underrepresented classes. Various modifications of SMOTE have been proposed to enhance its effectiveness.
The paper presents a novel Quantum-SMOTE algorithm that uses quantum processes for synthetic data generation. By applying clustering methods and rotation principles, the approach aims to improve minority class representation in datasets. The study evaluates the impact of Quantum-SMOTE on classification models using real-world telecom churn data.
Key metrics or figures:
Test accuracy improvement with 40% minority augmentation: 0.822183
F1 score enhancement with 50% minority augmentation: 0.834755
Stats
Test accuracy improvement with 40% minority augmentation: 0.822183
F1 score enhancement with 50% minority augmentation: 0.834755