Yapay Zeka Çerçevesi Üniversite Öğrencileri İçin İnovasyon Eğitimini Kişiselleştiriyor

Özet (abstract)

Deep Reinforcement Learning (DRL)-Powered Diagnosis and Customized Recommendation Model of College Students’ Innovation and Entrepreneurship Achievement; new guideline to personalized education assessment and recommendation in college. Conventional assessment techniques offer below standard battery tests of the learning curve, entrepreneurial participation, behavioural change and iteration of enhancement skills of college learners within the training course of action. This research introduces an adaptive intelligent framework, the Hybrid Transformer-Graph Reinforcement Adaptive Competency Engine (HT-GRACE), to model the achievement of students’ innovation- and entrepreneurship-related educational outcomes with proxy indicators from the Open University Learning Analytics Dataset (OULAD), consisting of 32,593 student records. The HT-GRACE framework integrates Graph Neural Networks (GNNs) for learning structural representations, Transformer Networks (TNs) for learning temporal behavioural representations, and Deep Reinforcement Learning using Proximal Policy Optimization (PPO) to simulate adaptive educational policy. Students’ conditions are represented by multiple learning interaction indices, and student intervention activities, including tutoring help, participation proposals for projects, innovation collaboration proposals, and adaptive learning enrichment modules, are adopted. The awards are formalized to measure educational advancement on proxy indicators and interventions steps was interpreted as simulated or anticipated suggestions. Evaluation shows high performance, with an Accuracy of 94.8% an F1-score of 94.86% a ROC-AUC of 94.95% and an MCC of 0.915. The framework also achieves a cumulative reward gain of 27.6% and a policy divergence stability score of 0.91, demonstrating stable performance in simulated adaptive educational decision-making. Applications include intelligent academic advising, entrepreneurial preparation platforms, competency-based analytics, and AI-driven personalized higher education ecosystems for diverse learners.

Yazarların özeti; kaynağından alınmıştır. Discover Artificial Intelligence, 2026 · DOI ↗