Research Interests

• Fitness landscape analysis
• Genetic programming
• Genetic algorithms
• Scheduling

Degrees

• PhD in Computing, Faculty of Electrical Engineering and Computing, University of Zagreb, Croatia, 2021.
• MSc in Mathematics, Department of Mathematics, University of Osijek,Croatia, 2014.
• BSc in Mathematics, Department of Mathematics, University of Osijek, Croatia, 2011.

Publications

1. M. Đurasević, M. Đumić, R. Čorić, F.J. Gil-Gala, Automated design of relocation rules for minimising energy consumption in the container relocation problem, Expert systems with applications 237 (2024)
   Abstract

   The container relocation problem (CRP) is a combinatorial optimisation problem in which the sequence of container relocations must be determined to retrieve all containers from the yard while optimising a given objective. Prior to now, the primary objective of CRP was to minimise the number of container relocations; however, due to environmental concerns, optimising energy consumption is gaining importance. This criterion was not considered extensively in the literature about CRP; therefore, there is a lack of suitable methods to tackle this problem variant. Primarily, there is a lack of relocation rules (RRs), simple heuristics that efficiently solve large problems in negligible time. Unfortunately, RRs are challenging to design manually since this process requires significant domain knowledge and is time-consuming. Using hyperheuristics to evolve new RRs automatically is one way to circumvent this problem. In this study, we evolved new RRs that aim to minimise energy consumption using genetic programming. We consider both the scenario in which only energy consumption is optimised and a multi-objective scenario where energy consumption is optimised together with the number of container relocations. The proposed approach is compared with an existing approach from the literature that uses a genetic algorithm to design RRs. The results show that RRs designed using genetic programming perform significantly better than the existing method, especially in multi-objective scenarios.
2. D. Ševerdija, R. Čorić, M. Orešković, L. Šošić, Detecting inflectional patterns for Croatian verb stems using class activation mappings, Journal of Language Modelling 12/1 (2024)
   Abstract

   All verbal forms in the Croatian language can be derived from two basic forms: the infinitive and the present stems. In this paper, we present a neural computation model which takes a verb in an infinitive form and finds a mapping to a present form. The same model can be applied vice-versa, i.e.~map a verb from present form to its infinitive form. Knowing the present form of a given verb one can deduce its inflections using grammatical rules. We experiment with our model on the Croatian language which belongs to the Slavic group of languages. In both classification tasks, the model learns a classifier and uses a class activation mapping to find characters in verbs that contribute to classification. We see that the model detects patterns that follow established grammatical rules for deriving present stem form from infinitive stem form and vice-versa.
3. R. Čorić, D. Matijević, D. Marković, PollenNet - a deep learning approach to predicting airborne pollen concentrations, Croatian Operational Research Review 14/1 (2023), 1-13
   Abstract

   The accurate short-term forecasting of daily airborne pollen concentrations is of great importance in public health. Various machine learning and statistical techniques have been employed to predict these concentrations. In this paper, an RNN-based method called PollenNet is introduced, which is capable of predicting the average daily pollen concentrations for three types of pollen: ragweed (Ambrosia), birch (Betula), and grass (Poaceae). Moreover, two strategies incorporating measurement errors during the training phase are introduced, making the method more robust. The data for experiments were obtained from the RealForAll project, where pollen concentrations were gathered using a Hirst-type 7-day volumetric spore trap. Additionally, five types of meteorological data were utilized as input variables. The results of our study demonstrate that the proposed method outperforms standard models typically used for predicting pollen concentrations, specifically the pollen calendar method, pollen predictions based on patterns, and the naive approach.
4. R. Čorić, M. Đumić, D. Jakobović, Genetic programming hyperheuristic parameter configuration using fitness landscape analysis, Applied Intelligence 51/10 (2021), 7402-7426
   Abstract

   Fitness landscape analysis is a tool that can help us gain insight into a problem, determine how hard it is to solve a problem using a given algorithm, choose an algorithm for solving a given problem, or choose good algorithm parameters for solving the problem. In this paper, fitness landscape analysis of hyperheuristics is used for clustering instances of three scheduling problems. After that, good parameters for tree-based genetic programming that can solve a given scheduling problem are calculated automatically for every cluster. Additionally, we introduce tree editing operators which help in the calculation of fitness landscape features in tree based genetic programming. A heuristic is proposed based on introduced operators, and it calculates the distance between any two trees. The results show that the proposed approach can obtain parameters that offer better performance compared to manual parameter selection.
5. M. Đumić, D. Šišejković, R. Čorić, D. Jakobović, Evolving priority rules for resource constrained project scheduling problem with genetic programming, Future Generation Computer Systems 86 (2018), 211-221
   Abstract

   The main task of scheduling is the allocation of limited resources to activities over time periods to optimize one or several criteria. The scheduling algorithms are devised mainly by the experts in the appropriate fields and evaluated over synthetic benchmarks or real-life problem instances. Since many variants of the same scheduling problem may appear in practice, and there are many scheduling algorithms to choose from, the task of designing or selecting an appropriate scheduling algorithm is far from trivial. Recently, hyper-heuristic approaches have been proven useful in many scheduling domains, where machine learning is applied to develop a customized scheduling method. This paper is concerned with the resource constrained project scheduling problem (RCPSP) and the development of scheduling heuristics based on Genetic programming (GP). The results show that this approach is a viable option when there is a need for a customized scheduling method in a dynamic environment, allowing the automated development of a suitable scheduling heuristic.
6. N. Čerkez, R. Čorić, M. Đumić, D. Matijević, Finding an optimal seating arrangement for employees traveling to an event, Croatian Operational Research Review 6/2 (2015), 419-427
   Abstract

   The paper deals with modelling a specific problem called the Optimal Seating Arrangement (OSA) as an Integer Linear Program and demonstrated that the problem can be efficiently solved by combining branch-and-bound and cutting plane methods. OSA refers to a specific scenario that could possibly happen in a corporative environment, i.e. when a company endeavors to minimize travel costs when employees travel to an organized event. Each employee is free to choose the time to travel to and from an event and it depends on personal reasons. The paper differentiates between using different travel possibilities in the OSA problem, such as using company assigned or a company owned vehicles, private vehicles or using public transport, if needed. Also, a user-friendly web application was made and is available to the public for testing purposes.

1. F.J. Gil-Gala, M. Đurasević, M. Đumić, R. Čorić, D. Jakobović, An analysis of training models to evolve heuristics for the travelling salesman problem, GECCO '23 Companion: Proceedings of the Companion Conference on Genetic and Evolutionary Computation, Lisbon, Portugal, 2023, 575-578
   Abstract

   Designing heuristics is an arduous task, usually approached with hyper-heuristic methods such as genetic programming (GP). In this setting, the goal of GP is to evolve new heuristics that generalise well, i.e., that work well on a large number of problems. To achieve this, GP must use a good training model to evolve new heuristics and also to evaluate their generalisation ability. For this reason, dozens of training models have been used in the literature. However, there is a lack of comparison between different models to determine their effectiveness, which makes it difficult to choose the right one. Therefore, in this paper, we compare different training models and evaluate their effectiveness. We consider the well-known Travelling Salesman Problem (TSP) as a case study to analyse the performance of different training models and gain insights about training models. Moreover, this research opens new directions for the future application of hyper-heuristics.
2. M. Đurasević, M. Đumić, R. Čorić, F.J. Gil-Gala, Automated design of relocation rules for minimising energy consumption in the container relocation problem, GECCO '23 Companion: Proceedings of the Companion Conference on Genetic and Evolutionary Computation, Lisbon, Portugal, 2023, 523-526
   Abstract

   The container relocation problem is a combinatorial optimisation problem aimed at finding a sequence of container relocations to retrieve all containers in a predetermined order by minimising a given objective. Relocation rules (RRs), which consist of a priority function and relocation scheme, are heuristics commonly used for solving the mentioned problem due to their flexibility and efficiency. Recently, in many real-world problems it is becoming increasingly important to consider energy consumption. However, for this variant no RRs exist and would need to be designed manually. One possibility to circumvent this issue is by applying hyperheuristics to automatically design new RRs. In this study we use genetic programming to obtain priority functions used in RRs whose goal is to minimise energy consumption. We compare the proposed approach with a genetic algorithm from the literature used to design the priority function. The results obtained demonstrate that the RRs designed by genetic programming achieve the best performance.
3. R. Čorić, M. Đumić, S. Jelić, A clustering model for time-series forecasting, 42nd International Convention - MIPRO 2019, Opatija, 2019, 1295-1299
   Abstract

   In this paper we consider a novel Integer programming approach for the cluster-based model used for time-series forecasting. There are several approaches in literature that aim to find a set of patterns which represent similar situations in the time series. In order to predict target variable, different types of fitting methods can be applied to set of data that belongs to the same pattern. We propose method that uses clustering of patterns and prediction of target value as the mean of values in the same cluster, in order to minimize total squared deviation between predicted and real values of target variable. We also propose a heuristic method that achieves good solution in practice. Our approach is applied to short-term prediction of airborne pollen concentrations. We give experimental results about comparison of our method to some common approaches.
4. R. Čorić, M. Đumić, S. Jelić, A Genetic Algorithm for Group Steiner Tree Problem, 41st International Convention - MIPRO 2018, Opatija, Hrvatska, 2018, 1113-1118
   Abstract

   In Group Steiner Tree Problem (GST) we are given a weighted undirected graph and family of subsets of vertices which are called groups. Our objective is to find a minimum-weight subgraph which contains at least one vertex from each group (groups do not have to be disjoint). GST is NP-hard combinatorial optimization problem that arises from many complex real-life problems such as finding substrate-reaction pathways in protein networks, progressive keyword search in relational databases, team formation in social networks, etc. Heuristic methods are extremely important for finding the good-enough solutions in short time. In this paper we present genetic algorithm for solving GST. We also give results of computational experiments with comparisons to optimal solutions.
5. R. Čorić, M. Đumić, D. Jakobović, Complexity Comparison of Integer Programming and Genetic Algorithms for Resource Constrained Scheduling Problems , 40th International ICT Convention - MIPRO 2017, Opatija, 2017, 1394-1400
   Abstract

   Resource constrained project scheduling problem (RCPSP) is one of the most intractable combinatorial optimization problems. RCPSP belongs to the class of NP hard problems. Integer Programming (IP) is one of the exact solving methods that can be used for solving RCPSP. IP formulation uses binary decision variables for generating a feasible solution and with different boundaries eliminates some of solutions to reduce the solution space size. All exact methods, including IP, search through entire solution space so they are impractical for very large problem instances. Due to the fact that exact methods are not applicable to all problem instances, many heuristic approaches are developed, such as genetic algorithms. In this paper we compare the time complexity of IP formulations and genetic algorithms when solving the RCPSP. In this paper we use two different solution representations for genetic algorithms, permutation vector and vector of floating point numbers. Two formulations of IP and and their time and convergence results are compared for the aforementioned approaches.
6. R. Čorić, S. Picek, D. Jakobović, C.A. Coello Coello, On the mutual information as a fitness landscape measure, GECCO '17 Proceedings of the Genetic and Evolutionary Computation Conference Companion, Berlin, Germany, 2017, 165-166
   Abstract

   Fitness landscape analysis plays an important role in both theoretical and practical perspectives when using evolutionary algorithms. In this paper, we develop a new measure based on the mutual information paradigm and we show how it can help to deduce further information about the fitness landscape. In order to validate it as a valuable source of information when conducting fitness landscape analysis, we investigate its properties on a well-known benchmark suite. Moreover, we investigate the usefulness of the obtained information when choosing crossover operators. Finally, we show that when using our new measure, a number of classifiers can be constructed that offer an improved accuracy.

Projects

• RZC PAN HR, (2022. – 2024.) with project leader Snježana Majstorović.
• Hyperheuristic Design of Dispatching Rules, funded by Croatian Science Foundation from 2020. Project leader: prof. Domagoj Jakobović from University of Zagreb.
• Application of optimization methods in biomedicine, bilateral project with Serbia, Duration: 01.01.2019. – 31.12. 2020. Project leaders: ass.prof. Slobodan Jelić from University of Osijek (croatian side) and ass.prof. Dušan Jakovetić from University of Novi Sad (serbian side).

Professional Activities

Conferences

• 15th International Conference on Operational Research KOI 2014, Osijek, Croatia, September 24-26, 2014.
• 40th International ICT Convention – MIPRO 2017, Opatija, Croatia, May 22-26, 2017.
• 41st International ICT Convention – MIPRO 2018, Opatija, Croatia, May 21-25, 2018.
• 42nd International ICT Convention – MIPRO 2019, Opatija, Croatia, May 20-24, 2019.
• 19th International Conference on Operational Research KOI 2022, Šibenik, Croatia, September 28-30, 2022.

Schools

• 7th PhD Summer School in Discrete Mathematics, Rogla, Slovenia, July 23-29, 2017.
• COST Training School: bridging the gap between theory and practice and making nature-inspired search and optimisation heuristics more applicable, Paris, France, October 18-24, 2017.
• International Workshop on Optimal Control of Dynamical Systems and Applications, Osijek, Croatia, June 20-22, 2018.
• Second Edition of the Summer School on Optimization, Big Data and Applications (OBA), Veroli, Italy, June 29th – July 6th, 2019.

Service Activities

• Festival znanosti:
  2011. radionica – Primjena Sunčeve svjetlosti pri određenim izračunavanjima
  2012. radionica – 10 u svijetu računala
  2013. radionica – Zamisli jedan broj
• Zimska škola matematike:
  2011. predavanje – Bertrandov paradoks
• Zimska škola informatike:
  2017. radionica – Multi-threading i multi-processing u Pythonu
• Geometrijska škola Stanka Bilinskog, Našice
  2017. radionica za osnovnu i srednju školu – Geometrijska vjerojatnost

Teaching

Nastavne aktivnosti u zimskom semestru akademske 2022./2023.

• Odjel za matematiku: Uvod u računalnu znanost, Heuristički algoritmi
• Prehrambeno tehnološki fakultet: Primijenjena matematika/Inženjerska matematika

Nastavne aktivnosti u ljetnom semestru akademske 2020./2021.

• Odjel za matematiku: Strukture podataka i algoritmi II, Moderni sustavi baza podataka, Obrada prirodnog jezika tehnikama dubinskog učenja

Prethodna nastava:

• Ekonomski fakultet: Matematika
• Odjel za matematiku: Uvod u računarstvo, Numerička analiza, Osnove umjetne inteligencije, Programiranje i programsko inženjerstvo
• Poljoprivredni fakultet Osijek, stručni studij u Vinkovcima: Matematika
• Poljoprivredni fakultet Osijek: Matematika (izvanredni studij)
• Prehrambeno tehnološki fakultet: Primijenjena matematika/Inženjerska matematika

Konzultacije (Office Hours):   Po dogovoru.