Status: Available

Title:

Description:

Status: Available

Title: Hyperparameter Optimization for Quantum-inspired Logic Decision Trees QLDTs

Description:

Quantum-Inspired Logic Decision Trees (QLDTs) provide an interpretable classification framework based on quantum logic. Unlike classical decision trees, QLDTs evaluate logical expressions over continuous-valued attributes, enabling more expressive decision boundaries. The predictive performance of QLDT depends on several parameters, including output thresholds, pruning criteria (if introduced), and weighting parameters. Currently, parameter selection is greatly manual. This proposal is on development a systematic hyper parameter optimization methods for QLDT/QLDT+. The goal is to answer the following research questions: 
Q1. Which parameters most strongly influence performance?
Q2. Which optimization algorithms are most effective?
Q3. Can automatic tuning improve generalization?

Many optimization techniques can be considered. The evaluation metrics on several datasets include metrics  such as accuracy, F1-score, Training time ...etc. 

Status: In Progress

Title: MC-QLDT: A Multi-class Quantum-inspired Logic Classifier
Description: Quantum‑inspired logic decision trees (QLDT) have been shown to effectively capture curved decision boundaries while maintaining interpretability. However, they are limited to binary classification tasks. In many practical application domains, such as medical diagnosis, data naturally span multiple classes. This work proposes MC‑QLDT, an extension that generalizes the QLDT framework to multi‑class problems. We investigate several strategies and we evaluate the resulting models on benchmark datasets against classical decision trees and random forests, with a focus on both predictive accuracy and interpretability. We expect this work to offer a practical and interpretable solution for multi‑class classification tasks where decision boundaries are non‑linear and transparent reasoning is essential for user trust and adoption.