Milan Đorđević

IEEE Conference Publication / IEEE Xplore

Goal-Based Multi-Agent AI Systems for Academic Advising: A Survey of Policies, Risk Detection, and Intervention Frameworks

Mohamad Helmi Klot; Milan Dordevic; George Tsaramirsis; Mohamad Nassereddine

https://ieeexplore.ieee.org/

Artificial intelligence (AI) is increasingly embedded in computer-based academic advising and decision-support systems to improve student retention, progression, and timely intervention at scale. However, most current platforms remain largely reactive, focusing on question answering or risk prediction, and lack the ability to explicitly model student goals, curriculum constraints, and adaptive intervention policies. This survey reviews goal-oriented, multi-agent, and agentic AI architectures for advising systems, with emphasis on software frameworks enabling policy-driven early-warning and intervention for at-risk students in higher education. It traces the evolution from rule-based degree audits and early-warning analytics to advanced architectures integrating learning analytics pipelines, explainable AI (XAI), affect-aware modelling, reinforcement learning, and LLM-based agents. A unified taxonomy classifies systems by goal representation, planning horizon, risk awareness, intervention autonomy, and governance. The survey highlights gaps between accurate risk detection and actionable intervention logic, examines closed-loop agentic architectures with human-in-the-loop controls, and outlines research challenges in causal validation, equity, scalability, and deployment.

Multi-agent Systems; Artificial Intelligence Systems; Intervention Framework; Academic Advisors; Policy Survey; Decision Support Systems; Explainable Artificial Intelligence; At-risk Students; Early Warning System; Learning Management System

—

IEEE Conference Publication / IEEE Xplore

Harnessing AI for Personalized Academic Major Recommendations An Application of Large Language Models in Education

Usman Durrani; Mustafa Akpinar; Asif Malik; Madeleine Togher; Milan Dordevic; Samer Aoudi

https://ieeexplore.ieee.org/

In the domain of educational counseling, the utilization of Large Language Models (LLMs) in conjunction with sophisticated AI technologies, such as embeddings and vector databases, introduces a groundbreaking methodology for advising students on their academic specializations. This study explores the complexities associated with conventional recommendation systems, which include issues such as data imbalance and insufficient contextual awareness. By adopting a Few-Shot Learning framework, the authors harness the flexibility of LLMs to identify critical factors related to students’ interests and competencies. The approach enables the dynamic extraction of important contextual information, thereby enhancing the predictive efficacy of the models.

Artificial Intelligence in Education; Large Language Models; Personalized Academic Advising; Educational Recommender Systems; AI-Powered Recommendations

—

IEEE Conference Publication / IEEE Xplore

Assessing the Effectiveness of Large Language Models in Predicting Student Dropout Rates

Usman Durrani; Mustafa Akpinar; Madeleine Togher; Asif Malik; Milan Dordevic; Samer Aoudi

https://ieeexplore.ieee.org/

Leveraging Large Language Models (LLMs) in educational analytics introduces promising and innovative approaches, particularly in predicting student dropout rates. This study investigates the integration of advanced artificial intelligence tools, including OpenAI LLMs, HuggingFace embeddings, and FAISS vector databases, within a Few-Shot Learning framework. Retrieval Augmented Generation (RAG) techniques were employed to improve the quality of contextual data provided to these models. The research highlights the potential of LLMs as tools for dropout prediction in academic settings.

Large Language Models; Student Dropout Forecasting; Few-Shot Learning; AI in Education; Contextual Data Analysis; Recommender Systems

—

Journal of Infrastructure, Policy and Development

Design, development, and evaluation of a mobile application for safety engineers

Suat Kasap; Ersin Elbasi; Milan Dordevic

https://doi.org/10.24294/jipd.v8i12.8107

Hazards are the primary cause of occupational accidents, as well as occupational safety and health issues. Therefore, identifying potential hazards is critical to reducing the consequences of accidents. This paper explains the processes involved in developing a mobile application designed for hazard analysis for safety engineers, following the Unified Modeling Language (UML) approach. The mobile application was developed through a 10-step approach and supports data collection through voice notes, photos and videos, enabling a more convenient alternative to manual note-taking.

safety engineering; occupational safety and health; mobile application; unified modeling language (UML); occupational safety and health policy; risk assessment; hazard analysis

—

IEEE Conference Publication / IEEE Xplore

An Accurate Brain Tumor Segmentation using Deep Learning

Asad Safi; Milan Dordevic; Reem Atassi; Fuad AlHosban

https://ieeexplore.ieee.org/

Gliomas, which can appear in different sizes, locations and with scattered boundaries, pose a significant challenge as an aggressive type of tumor. Convolutional Neural Networks (CNNs) were used to develop an automatic deep learning-based 2D-CNN model for brain tumor segmentation. The model architecture uses small convolution filters and an increased number of convolutional layers. The proposed method was evaluated using the BRATS-2015 dataset and the Dice Similarity Coefficient (DSC), achieving results comparable to other 2D CNN architecture-based methods.

Deep Learning; Neural Network; Image Analysis

—

International Journal of Transport Development and Integration

AI-Enabled Assessment of Roadway Integrity: Forecasting Bitumen Deformation and Road Stability Throughout the Lifecycle Under Traffic Impact

Sabahudin Vrtagic; Milan Dordevic; Fatih Dogan; Muhammed Codur; Mario Hoxha; Edis Softic

https://doi.org/10.18280/ijtdi.070406

Asphalt-paved road junctions frequently encounter deformation and degradation challenges due to heavy vehicular traffic and varying climatic conditions. This study employs a multifaceted approach, incorporating a Multilayer Perceptron (MLP) model, ancillary machine learning techniques, and optimization methodologies. The findings highlight the contribution of heavy vehicles to road erosion and the impact of vehicular retention and braking at intersections. The study presents a comprehensive approach to sustainable and efficient road infrastructure management.

Asphalt pavements; Backpropagation; Deformation; Intersections; Machine learning; Predictive models; Traffic loads

—

Axioms

Development of Integrated Linear Programming Fuzzy-Rough MCDM Model for Production Optimization

Milan Dordevic; Rade Tešić; Srdjan Todorović; Miloš Jokić; Dillip Kumar Das; Željko Stević; Sabahudin Vrtagic

https://doi.org/10.3390/axioms11100510

One of the most common tools for achieving optimization and adequate production process management is linear programming (LP). This paper integrates linear programming and a Multi-Criteria Decision-Making (MCDM) model. Linear programming was first applied to optimize production and identify several potential solutions, followed by evaluation using the Improved Fuzzy Stepwise Weight Assessment Ratio Analysis (IMF SWARA). A novel Rough CRADIS method was developed and verified through comparative analysis.

linear programming; IMF SWARA; production optimization; Rough CRADIS

—

Fusion: Practice and Applications

A New Data Fusion Model for Medical Image Encryption in IoT Environment

Reem Atassi; Fuad Alhosban; Milan Dordevic

https://doi.org/10.54216/FPA.080102

The improvement of the Internet of Things (IoT) is expected to transform the healthcare industry and support the rise of the Internet of Medical Things (IoMT). This study introduces a new data fusion model in an IoT environment. The proposed SSOECC-MIC model focuses on an effective encryption scheme with an optimal key generation process for medical image protection, using ECC encryption and swallow swarm optimization.

Security; Data Fusion; Internet of Things; Healthcare; Medical images; Encryption; Key generation

—

Axioms

Ranking road sections based on MCDM model: New improved fuzzy SWARA (IMF SWARA)

Sabahudin Vrtagić; Edis Softić; Marko Subotić; Željko Stević; Milan Dordevic; Mirza Ponjavic

https://www.mdpi.com/2075-1680/10/2/92

Traffic management is a difficult and demanding task. Through this paper, an integrated fuzzy model for ranking road sections based on four inputs and four outputs was developed. The contribution of the paper is reflected in the development of the improved fuzzy step-wise weight assessment ratio analysis (IMF SWARA) method and its integration with the fuzzy MARCOS method. The obtained results were verified through a three-phase sensitivity analysis, and the stability of the model was proven.

Road Section; IMF SWARA; traffic safety; fuzzy MARCOS; DEA

—