Item recommendation and quantum correlation on multiple datasets

Area of Science:

• Machine Learning
• Quantum Computing
• Recommender Systems

Background:

• Recommender systems are vital for personalized user experiences in e-commerce and entertainment.
• Traditional correlation methods struggle with complex, high-dimensional data.
• Quantum computing offers a novel approach to enhance correlation computations.

Purpose of the Study:

• To compare the efficiency of classical and quantum correlation techniques in recommender systems.
• To evaluate the Item Recommendation and Quantum Correlation (IRQC) method.
• To analyze performance across four diverse datasets: Supermarket Sales, IMDB Top 250 movies, MovieLens 10k, and BigBasket products.

Main Methods:

• Utilized parameterized quantum circuits with rotation gates and entanglement for the IRQC method.
• Employed both classical and quantum correlation approaches for comparative analysis.
• Evaluated efficacy using Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE).

Main Results:

• Quantum correlation techniques consistently outperformed classical methods across all tested datasets.
• The Quantum Correlation approach achieved lower MAE values (e.g., 0.30 for Supermarket Sales) compared to classical methods (e.g., 1.48 for Supermarket Sales).
• Significant improvements in recommendation accuracy were observed using quantum methods.

Conclusions:

• Quantum computing holds significant potential for advancing machine learning applications.
• The proposed Quantum Correlation approach demonstrates superior performance in enhancing recommender systems.
• Quantum methods offer a promising direction for more precise and efficient personalized recommendations.