{"id":52272,"date":"2025-12-18T12:04:46","date_gmt":"2025-12-18T11:04:46","guid":{"rendered":"https:\/\/www.mathos.unios.hr\/optimaai\/"},"modified":"2026-10-29T16:05:55","modified_gmt":"2026-10-29T15:05:55","slug":"optimaai","status":"publish","type":"page","link":"https:\/\/www.mathos.unios.hr\/en\/optimaai\/","title":{"rendered":"Advanced Algorithms and Optimization Models Supported by Mathematical Theory (OptimaAI)"},"content":{"rendered":"\n
\n
\n
\"OptimaAI<\/figure>\n\n\n\n

Basic Information<\/h2>\n\n\n\n

The OptimaAI project focuses on the development of innovative artificial intelligence algorithms and optimization models grounded in modern mathematical theories. The aim of the project is to build energy-efficient and scalable solutions for the analysis of complex data in bioinformatics, natural language processing, visual tracking, and industrial optimization. A key component of the project is the strong synergy between applied AI technologies (deep learning, NLP, edge AI) and fundamental mathematical disciplines (combinatorics and discrete mathematics, number theory, and group theory).<\/p>\n\n\n\n

The project is structured around four work packages: WP1 \u2013 Development of advanced bioinformatics tools for single-cell transcriptome analysis; WP2 \u2013 Development of AI models for large-scale datasets and intelligent real-time camera signal processing; WP3 \u2013 Deep learning for combinatorial optimization and graph theory; and WP4 \u2013 Solving mathematical problems applicable to optimization through heuristics and metaheuristics using theoretical, combinatorial, and discrete approaches.<\/p>\n\n\n\n

Expected results include open-source software, high-impact scientific publications, and the transfer of results into applications relevant to healthcare, industry, and the digital transition.<\/p>\n\n\n\n\n\n
Full title<\/th>Advanced Algorithms and Optimization Models Supported by Mathematical Theory \u2013 OptimaAI<\/td><\/tr>\n
Project type<\/th>Institutional research project of the Josip Juraj Strossmayer University of Osijek<\/td><\/tr>\n
Funding source<\/th>National Recovery and Resilience Plan 2021\u20132026 (source 581 \u2013 Recovery and Resilience Facility), Programme Agreement between the Ministry of Science, Education and Youth and the University of Osijek<\/td><\/tr>\n
Contract number<\/th>581-UNIOS-54<\/td><\/tr>\n
Beneficiary<\/th>School of Applied Mathematics and Informatics, Josip Juraj Strossmayer University of Osijek<\/td><\/tr>\n
Principal investigator<\/th>Assoc. Prof. Domagoj Matijevi\u0107, PhD<\/td><\/tr>\n
Project value<\/th>EUR 194,999.70 (100 % grant)<\/td><\/tr>\n
Duration<\/th>1 October 2025 \u2013 30 September 2029 (48 months)<\/td><\/tr>\n<\/tbody>\n<\/table>\n\n\n\n

Publications<\/h2>\n\n\n\n

Journal papers<\/h3>\n\n\n\n
    \n
  1. I. Mati\u0107, Zelevinsky segment and strongly positive representations, Canadian Mathematical Bulletin<\/em> (2026), accepted. doi:10.4153\/S0008439526102781<\/a><\/li>\n\n\n\n
  2. T. Prusina, J. Beni\u0107, D. Matijevi\u0107, D. \u0160everdija, Multi-person 2D human pose estimation: a benchmark for real-time applications, Engineering Review<\/em> 46 (2026). Presented at MOTSP 2026. Code: HPEBench2D<\/a><\/li>\n\n\n\n
  3. J. Maltar, I. Markovi\u0107, D. Matijevi\u0107, I. Petrovi\u0107, Deep recurrent visual place recognition based on softmax fine-tuned image representations, Engineering Review<\/em> 46 (2026). Presented at MOTSP 2026.<\/li>\n\n\n\n
  4. M. Juki\u0107 Bokun, I. Soldo, Triangular D(\u22121)-tuples, Bull. Math. Soc. Sci. Math. Roumanie<\/em> 69\/2 (2026), 233\u2013239.<\/li>\n\n\n\n
  5. D. Krupi\u0107, D. Matijevi\u0107, N. \u0160uvak, J. Maltar, D. \u0160everdija, Evaluating the agreement between human preferences, GPT-4V and Gemini Pro Vision assessments: Can AI recognize what people might like?<\/a>, Computers in Human Behavior: Artificial Humans<\/em> 6 (2025)<\/li>\n<\/ol>\n\n\n\n

    Conference proceedings papers<\/h3>\n\n\n\n
      \n
    1. J. Sabljo, M. \u0110umi\u0107, M. \u0110urasevi\u0107, Evolving Dispatching Rules for the Unrelated Parallel Machines Scheduling Problem with Precedence and Resource Constraints with Genetic Programming, GECCO ’26 \u2013 Genetic and Evolutionary Computation Conference<\/em>, San Jose, Costa Rica, 13\u201317 July 2026, ACM. doi:10.1145\/3795095.3805192<\/a><\/li>\n\n\n\n
    2. J. Sabljo, M. \u0110umi\u0107, M. \u0110urasevi\u0107, Analysis of Training Models for Evolving Heuristics in the Unrelated Parallel Machines Scheduling Problem with Additional Resources, 2026 MIPRO 49th ICT and Electronics Convention<\/em>, Opatija, Croatia, 2026. Code: UPMSP-PR<\/a><\/li>\n\n\n\n
    3. L. Borozan, B. Borozan, D. \u0160everdija, Efficient Checkpointing via Object Serialization in C# Applications, 2026 MIPRO 49th ICT and Electronics Convention<\/em>, Opatija, Croatia, 2026. Code: CheckpointingSystem<\/a><\/li>\n<\/ol>\n\n\n\n

      Papers under review<\/h3>\n\n\n\n
        \n
      1. K. Sabo, R. Scitovski, D. Grahovac, \u0160. Ungar, Mahalanobis clustering for color image segmentation and application to skin lesion classification, Advances in Data Analysis and Classification<\/em> (submitted)<\/li>\n\n\n\n
      2. K. K. Gottwald, T. Hofmann, S. Majstorovi\u0107 Ergoti\u0107, On the total inverse flow of a graph, Aequationes Mathematicae<\/em> (submitted 2026)<\/li>\n\n\n\n
      3. S. Majstorovi\u0107 Ergoti\u0107, K. K. Gottwald, The total inverse flow of a graph: A \u0160olt\u00e9s-type problem, Bulletin of the Malaysian Mathematical Sciences Society<\/em> (submitted 2026)<\/li>\n\n\n\n
      4. L. Borozan, D. Matijevi\u0107, Scenario-Driven Control of Real-Time LLM Agents, MIPRO Robotics 2026 \u2013 5th Robotics Conference (RTA)<\/em>, Zadar, Croatia (accepted, proceedings forthcoming)<\/li>\n<\/ol>\n\n\n\n

        All papers acknowledge the OptimaAI project (581-UNIOS-54), funded by the National Recovery and Resilience Plan.<\/em><\/p>\n\n\n\n

        Conferences and workshops<\/h2>\n\n\n\n\n\n
        Date<\/th>Event<\/th>Participation<\/th><\/tr><\/thead>\n
        25\u201328 Sep 2026<\/td>GCB 2026 \u2013 German Conference on Bioinformatics<\/td>B. Borozan, poster \u201cCombinatorial CAR T-cell Circuit Design using Single- and Multi-Objective Local Search\u201d<\/td><\/tr>\n
        13\u201317 Jul 2026<\/td>GECCO 2026, San Jose, Costa Rica<\/td>J. Sabljo, paper presentation (Sabljo, \u0110umi\u0107, \u0110urasevi\u0107); M. \u0110umi\u0107, attendance<\/td><\/tr>\n
        10\u201312 Jul 2026<\/td>GECCO 2026 SIGEVO Summer School, San Jose<\/td>J. Sabljo<\/td><\/tr>\n
        2026<\/td>MIPRO Robotics 2026 \u2013 5th Robotics Conference (RTA), Zadar, Croatia<\/td>L. Borozan, paper presentation (Borozan, Matijevi\u0107)<\/td><\/tr>\n
        26\u201328 May 2026<\/td>MOTSP 2026 \u2013 Management of Technology Step to Sustainable Production, Zadar, Croatia<\/td>talks: J. Maltar (visual place recognition), J. Beni\u0107 (2D pose estimation benchmark), T. Prusina (edge-based bearing condition monitoring); D. \u0160everdija, attendance<\/td><\/tr>\n
        25\u201329 May 2026<\/td>MIPRO 2026 \u2013 49th ICT and Electronics Convention, Opatija, Croatia<\/td>talks: J. Sabljo (training models for evolving heuristics), L. Borozan (efficient checkpointing)<\/td><\/tr>\n
        26 May 2026<\/td>Project workshop \u201cHyper-heuristic methods for electric vehicle routing problems\u201d, FER, University of Zagreb<\/td>M. \u0110umi\u0107, R. \u010cori\u0107, J. Sabljo<\/td><\/tr>\n<\/tbody>\n<\/table>\n\n\n\n

        Open-source code repositories<\/h2>\n\n\n\n\n\n
        Repository<\/th>Content<\/th>Work package<\/th><\/tr><\/thead>\n
        HPEBench2D<\/a><\/td>real-time multi-person 2D pose estimation benchmark<\/td>WP2<\/td><\/tr>\n
        CheckpointingSystem<\/a><\/td>efficient checkpointing via object serialization in C# applications<\/td>WP3<\/td><\/tr>\n
        UPMSP-PR<\/a><\/td>genetic programming for unrelated parallel machine scheduling<\/td>WP4<\/td><\/tr>\n<\/tbody>\n<\/table>\n\n\n\n

        Collaborations and mobility<\/h2>\n\n\n\n\n\n
        Institution<\/th>Collaboration<\/th>Period<\/th><\/tr><\/thead>\n
        University of Regensburg, Faculty of Informatics and Data Science (Germany)<\/td>research stay of Bartol Borozan; joint work on marker gene selection (WP1)<\/td>1 Oct 2025 \u2013 30 Jun 2026<\/td><\/tr>\n
        University of Regensburg, Faculty of Informatics and Data Science (Germany)<\/td>short research visit of Luka Borozan (Erasmus+)<\/td>24\u201328 Mar 2026<\/td><\/tr>\n
        Chemnitz University of Technology, Faculty of Mathematics (Germany)<\/td>S. Majstorovi\u0107 Ergoti\u0107 is scientific co-supervisor of PhD candidate Kurt Klement Gottwald; two joint papers (WP3)<\/td>since 2026<\/td><\/tr>\n
        Faculty of Electrical Engineering and Computing, University of Zagreb<\/td>joint papers with M. \u0110urasevi\u0107 (WP4); project workshop, 26 May 2026<\/td>2026<\/td><\/tr>\n
        GECCO 2026 SIGEVO Summer School, San Jose (Costa Rica)<\/td>training of Josipa Sabljo<\/td>10\u201312 Jul 2026<\/td><\/tr>\n<\/tbody>\n<\/table>\n\n\n\n

        Research infrastructure<\/h2>\n\n\n\n

        The project procured low-power edge AI devices (November 2025), which are used to validate compressed deep-learning models and to measure speed and resource consumption on constrained hardware (WP2, WP3).<\/p>\n\n\n\n

        In the second project year a next-generation GPU cluster is being procured; it will be integrated into the School\u2019s HPC infrastructure and shared by all work packages (public procurement launched in October 2026). Equipment purchased with project funds is labelled with the EU emblem and funding statement.<\/p>\n\n\n\n

        News<\/h2>\n\n\n\n