Congratulations Diksha for the acceptance of your article titled Quantum-inspired Attribute Selection Algorithms in Quantum Science and Technology - Great Work, many more to come!

Diksha Sharma et al., Quantum Sci. Technol. 10, 015036 (2025)

In this work, we have proposed and analyzed quantum splitting criteria based on quantum information gain and fidelity to construct a quantum decision tree. For the quantum information gain-based criterion, we have utilized angle embedding to construct a quantum state out of the classical dataset, which is further used to compute quantum mutual information between a feature and the class attribute. For the fidelity-based criterion, we have efficiently utilized the probability distribution to generate a quantum state, which is further used to compute fidelity for each feature. The numerical analysis showed that the fidelity splitting criterion selects the feature with a uniform probability distribution. This further assists in obtaining a balanced and more accurate decision tree. Our results demonstrated that the proposed QIG and fidelity-based criteria can provide a significant improvement in terms of different evaluation metrics for imbalanced and balanced datasets. For a comprehensive analysis, we have examined the efficiency of all quantum and classical splitting methods on specificity, positive predictive value, and negative predictive value, which play a crucial role in medical datasets. In particular, our fidelity-based criterion significantly succeeds all classical splitting criteria for all evaluation metrics considered in this study, for all balanced and imbalanced datasets including the larger ones. The obtained results clearly demonstrate the advantages of quantum splitting criteria over classical splitting criteria.