Amirkabir University of Technology AUT Journal of Modeling and Simulation 2588-2953 58 1 2026 06 01 A Stability-Guided Quantum Feature Selection and Quantum Kernel SVM Framework for Entrepreneurial Competency Prediction 6053 10.22060/miscj.2026.25629.5484 EN Mohammad Reza Mehregan Prof., Department of Industrial Management, Faculty of Management, University of Tehran, Tehran, Iran. 0000-0002-7974-8171 Arman Rezasoltani Ph.D. Candidate, Department of Industrial Management, Faculty of Management, University of Tehran, Tehran, Iran. 0009-0003-6960-6713 Amir Mohammad Khani Ph.D. Candidate, Department of Industrial Management, Faculty of Management, University of Tehran, Tehran, Iran. 0000-0001-8798-2956 Journal Article 2026 03 16 The key challenges of performance prediction in entrepreneurship stem from the heterogeneous, multidimensional, nonlinear, and imbalanced characteristics of entrepreneurial behaviors, i.e., ambitiousness and risk preference. While machine learning algorithms, including Support Vector Machines, Random Forests, XGBoost, and LightGBM, have been purported and adopted to model entrepreneurial outcomes, their optimization saturation effect in performance prediction between 60% and 79% is observed, with little novelty in representational expansion by resampling and ensemble engineering approaches. To overcome the limits of existing models, this paper adopted a stability-guided quantum-enhanced hierarchical QAOA-QSVM framework whereby hierarchical Quantum Approximate Optimization Algorithm (QAOA) feature selection interacted with Quantum Kernel Support Vector Machine for Entrepreneurial Skill (QSVM). Features extracted from 219 university students' gender, race, personality, and university attributes were successively selected by formulating the QUBO problem of feature optimization and solving with QAOA. The stability consensus was weighed by stratified five-folds to improve reproducibility and robustness. Selected features were eventually embedded in an entangled quantum Hilbert space of the ZZFeatureMap to enable high-order nonlinear feature interactions with quantum kernel learners. The results indicated that our quantum approach surpassed a classical RBF-SVM on information from university students by nearly 22% (83.09% vs. 60.68%) and reached an 86.84% recall for entrepreneur prediction. We conclude that quantum representation learning structures can provide more insightful feature embedding ways, enabling novel transformers in talent pipeline building of entrepreneurship courses. This study also extends a meta-complementary theory between quantum machine learning and entrepreneurship.

Amirkabir University of Technology AUT Journal of Modeling and Simulation 2588-2953 58 1 2026 06 01 Neuro-Symbolic Claim Detection for Arabic: Integrating Graph Neural Networks with Efficiently Fine-Tuned Small Language Models 5917 10.22060/miscj.2025.24744.5431 EN Dina Falah Masood Alborz Campus, University of Tehran, Tehran, Iran 0009-0003-1922-0652 Heshaam Faili College of Engineering, School of Electrical and Computer Engineering, University of Tehran, Tehran, Iran 0000-0003-0443-3762 Journal Article 2025 09 16 The rise in online information has increased the spread of misinformation and disinformation, which makes claim detection important for the reliability of digital content. Claim detection refers to the problem of determining whether a factual statement is real or fake. While there has been great advancement in high-resource languages, Arabic is particularly challenging due to its special characteristics and the low availability of annotated resources. In this paper, we make use of the Arabic News Stance (ANS) corpus and develop a multi-stage framework to help tackle these problems. Our method effectively models the structural interplay between claims and words with a carefully heterogeneous bipartite graph and the use of advanced Graph Neural Networks (GNNs). In parallel, we used two Small-scale Language Models (SLMs) in this domain, namely LLaMA-3.2-1B and Qwen2.5-3B, and demonstrated that together with QLoRA-based fine-tuning, these lightweight models can be as effective as or even better than heavier baselines. To this point, we suggested a novel neuro-symbolic fusion paradigm that unifies structural reasoning with linguistic prior information. By integrating the outputs from a Graph Convolutional Network (GCN) with the fine-tuned Qwen2.5-3B logits, we achieved the state-of-the-art performance on the ANS dataset with an accuracy of 0.732 and an F1-score of 0.717. This framework shows that the combined results of structural graph reasoning and SLMs lead to accurate and computationally efficient claim detection. This work not only enhances the detection of Arabic misinformation but also provides a scalable solution for low-resource languages.