TY - JOUR AU - Kegyes, Tamás AU - Kummer, Alex AU - Süle, Zoltán AU - Abonyi, János TI - Generally Applicable Q-Table Compression Method and Its Application for Constrained Stochastic Graph Traversal Optimization Problems JF - INFORMATION (BASEL) J2 - INFORMATION-BASEL VL - 15 PY - 2024 IS - 4 SP - 193 SN - 2078-2489 DO - 10.3390/info15040193 UR - https://m2.mtmt.hu/api/publication/34762966 ID - 34762966 AB - We analyzed a special class of graph traversal problems, where the distances are stochastic, and the agent is restricted to take a limited range in one go. We showed that both constrained shortest Hamiltonian pathfinding problems and disassembly line balancing problems belong to the class of constrained shortest pathfinding problems, which can be represented as mixed-integer optimization problems. Reinforcement learning (RL) methods have proven their efficiency in multiple complex problems. However, researchers concluded that the learning time increases radically by growing the state- and action spaces. In continuous cases, approximation techniques are used, but these methods have several limitations in mixed-integer searching spaces. We present the Q-table compression method as a multistep method with dimension reduction, state fusion, and space compression techniques that project a mixed-integer optimization problem into a discrete one. The RL agent is then trained using an extended Q-value-based method to deliver a human-interpretable model for optimal action selection. Our approach was tested in selected constrained stochastic graph traversal use cases, and comparative results are shown to the simple grid-based discretization method. LA - English DB - MTMT ER - TY - JOUR AU - Kegyes, Tamás AU - Süle, Zoltán AU - Abonyi, János TI - Disassembly line optimization with reinforcement learning JF - CENTRAL EUROPEAN JOURNAL OF OPERATIONS RESEARCH J2 - CEJOR VL - 2024 PY - 2024 SP - 1 SN - 1435-246X DO - 10.1007/s10100-024-00906-3 UR - https://m2.mtmt.hu/api/publication/34728824 ID - 34728824 AB - As the environmental aspects become increasingly important, the disassembly problems have become the researcher’s focus. Multiple criteria do not enable finding a general optimization method for the topic, but some heuristics and classical formulations provide effective solutions. By highlighting that disassembly problems are not the straight inverses of assembly problems and the conditions are not standard, disassembly optimization solutions require human control and supervision. Considering that Reinforcement learning (RL) methods can successfully solve complex optimization problems, we developed an RL-based solution for a fully formalized disassembly problem. There were known successful implementations of RL-based optimizers. But we integrated a novel heuristic to target a dynamically pre-filtered action space for the RL agent ( dl O pt RL algorithm) and hence significantly raise the efficiency of the learning path. Our algorithm belongs to the Heuristically Accelerated Reinforcement Learning (HARL) method class. We demonstrated its applicability in two use cases, but our approach can also be easily adapted for other problem types. Our article gives a detailed overview of disassembly problems and their formulation, the general RL framework and especially Q-learning techniques, and a perfect example of extending RL learning with a built-in heuristic. LA - English DB - MTMT ER - TY - JOUR AU - Kegyes, Tamás AU - Süle, Zoltán AU - Abonyi, János TI - Machine learning -based decision support framework for CBRN protection JF - HELIYON J2 - HELIYON VL - 10 PY - 2024 IS - 2 SN - 2405-8440 DO - 10.1016/j.heliyon.2024.e25946 UR - https://m2.mtmt.hu/api/publication/34570694 ID - 34570694 LA - English DB - MTMT ER - TY - JOUR AU - Kegyes, Tamás AU - Süle, Zoltán AU - Abonyi, János TI - Az információmenedzsment szerepe az ABV-védelemben JF - NEMZETBIZTONSÁGI SZEMLE (ONLINE) J2 - NEMZETBIZT SZLE VL - 11 PY - 2023 IS - 1 SP - 62 EP - 77 PG - 16 SN - 2064-3756 DO - 10.32561/nsz.2023.1.5 UR - https://m2.mtmt.hu/api/publication/33832252 ID - 33832252 AB - Az atom-, bio- és vegyi (ABV-) incidensek felderítése kiemelt fontosságú feladat, amely évtizedek óta intenzíven kutatott téma. A folyamatos technológiai, adatfeldolgozási és automatizálási vívmányok újabb és újabb fejlesztési potenciált nyitnak az ABV-védelem terén is, amely napjainkra komplex, interdiszciplináris tudományterületté vált. Ennek megfelelően kémikusok, fizikusok, meteorológusok, katonai szakértők, programozók és adattudósok egyaránt közreműködnek a kutatásokban. A hazai ABV-védelmi képességek hatékony növelésének a kulcsa is abban rejlik, hogy megfelelően strukturált koncepció mentén folyamatos és célirányos fejlesztés történjen. Kutatásunk célja, hogy áttekintést adjunk a modern ABV-védelmi technológiák főbb komponenseiről, ezen belül összefoglaljuk az ABV-felderítés, illetve a döntéstámogatási lépések koncepcionális követelményeit, és bemutatjuk az információmenedzsment szerepét és legújabb lehetőségeit a folyamatokban. LA - Hungarian DB - MTMT ER - TY - JOUR AU - Süle, Zoltán AU - Baumgartner, János AU - Leitold, Dániel AU - Dulai, Tibor AU - Orosz, Ákos AU - Fogarassyné Vathy, Ágnes TI - Software Framework and Graph-based Methodology for Optimal Patient Appointment Planning JF - PROCEDIA COMPUTER SCIENCE J2 - PROC COMPUTER SCI VL - 219 PY - 2023 SP - 1169 EP - 1176 PG - 8 SN - 1877-0509 DO - 10.1016/j.procs.2023.01.398 UR - https://m2.mtmt.hu/api/publication/33714031 ID - 33714031 LA - English DB - MTMT ER - TY - CHAP AU - Kegyes, Tamás AU - Süle, Zoltán AU - Abonyi, János ED - Pintér, Miklós TI - Incorporation of heuristic search into Q-learning T2 - Short papers of VOCAL 2022 (9th VOCAL Optimization Conference: Advanced Algorithms) PB - Magyar Operációkutatási Társaság CY - Budapest SN - 9786150159874 PY - 2022 UR - https://m2.mtmt.hu/api/publication/33067088 ID - 33067088 AB - In the last years, Reinforcement Learning (RL) methods have proven its efficiency on multiple complex problems. For discrete problems Q-learning methods delivered an easily implementable solution. In this context a major challenge is to find a proper ratio between exploiting the current knowledge and exploring unknown patterns. For exploration, usually a purely random action taking mechanism is applied, which leads to a slow convergence to an optimal solution. In our presentation we will take an overview how different heuristics were built-in into RL methods to improve their learning efficiency, and we will present a categorization of the different approaches. Then we will demonstrate the effect of incorporating selected heuristic into a standard Q-learning based solution on a disassembly line optimisation problem. LA - English DB - MTMT ER - TY - JOUR AU - Bakon, Krisztián Attila AU - Holczinger, Tibor AU - Süle, Zoltán AU - Jaskó, Szilárd AU - Abonyi, János TI - Scheduling Under Uncertainty for Industry 4.0 and 5.0 JF - IEEE ACCESS J2 - IEEE ACCESS VL - 10 PY - 2022 SP - 74977 EP - 75017 PG - 41 SN - 2169-3536 DO - 10.1109/ACCESS.2022.3191426 UR - https://m2.mtmt.hu/api/publication/33020481 ID - 33020481 N1 - cited By 0 LA - English DB - MTMT ER - TY - JOUR AU - Kegyes, Tamás AU - Süle, Zoltán AU - Abonyi, János TI - The Applicability of Reinforcement Learning Methods in the Development of Industry 4.0 Applications JF - COMPLEXITY J2 - COMPLEXITY VL - 2021 PY - 2021 SP - 1 EP - 31 PG - 31 SN - 1076-2787 DO - 10.1155/2021/7179374 UR - https://m2.mtmt.hu/api/publication/32517499 ID - 32517499 N1 - cited By 0 LA - English DB - MTMT ER - TY - CONF AU - Kegyes, Tamás AU - Süle, Zoltán AU - Abonyi, János ED - Sziklai, Balázs TI - Szétszerelési láncok optimálisa megerősítéses tanulással T2 - XXXIV. Magyar Operációkutatási Konferencia: Absztraktok könyve PB - Gazdaságmodellezési Társaság C1 - Budapest PY - 2021 SP - 111 UR - https://m2.mtmt.hu/api/publication/32176984 ID - 32176984 LA - Hungarian DB - MTMT ER - TY - CONF AU - Süle, Zoltán AU - Baumgartner, János AU - Fogarassyné Vathy, Ágnes ED - Sziklai, Balázs TI - Betegellátási folyamatok optimális ütemezése T2 - XXXIV. Magyar Operációkutatási Konferencia: Absztraktok könyve PB - Gazdaságmodellezési Társaság C1 - Budapest PY - 2021 SP - 61 UR - https://m2.mtmt.hu/api/publication/32176983 ID - 32176983 LA - Hungarian DB - MTMT ER -