Affiliations:


Asst. Prof. Marko Đurasević, Ph.D.


Ph.D. in Computer Science

Department of Electronics, Microelectronics, Computer and Intelligent Systems (ZEMRIS)

University of Zagreb, Faculty of Electrical Engineering and Computing (FER)

Unska 3, 10000 Zagreb, Croatia


Contact:


E-mail: marko.durasevic[at]fer.hr


Room: D338


Possible research areas and topics:


  1. Evolutionary computing - implementation or application or different evolutionary computation algorithms

  2. Hyperheuristics - methods for automatic design of new heuristics for different combinatorial optimisation problem
    • https://en.wikipedia.org/wiki/Hyper-heuristic
    • Genetic programming, gene expression programming, Cartesian genetic programming
    • Improving their performance and execution time:
      • Surrogate models
      • Local search
      • Ensemble learning
      • Feature selection and construction

  3. Scheduling problems - solving different scheduling problem variants
    • In general, a set of jobs needs to be scheduled on a set of machines that can execute them
    • Applying different metaheuristic methods
    • Unrelated machines scheduling
      • Batch scheduling - machines can execute several jobs simultaneously
      • Resource constraints - jobs consume additional resources while being executed
      • Green scheduling - reduce energy consumption of the generatad schedules
    • Resource constrained project scheduling problem
    • Timetabling - creating exam and lecture schedules

  4. Vehicle routing problems - solving different problems where vehicles need to be routed towards customers
    • https://en.wikipedia.org/wiki/Vehicle_routing_problem
    • https://neo.lcc.uma.es/vrp/
    • Applying different metaheuristic methods: genetic algorithms, genetic programming, etc.
    • Various problem variants:
      • VRP with time windows - customers can only be served during a certain time period
      • Pickup and delivery - items need to be picked up at some customers and delivered to others
      • VPR with profits - customers can be rejected in order to improve the profit
    • Green vehicle routing problem - problems considering influence on the environment (using electric vehicles, reducing emissions, etc.)
    • Dynamic vehicle routing problem - routing problems in which certain parameters change over time
      • New customers arrive over time
      • Customer properties are subject to change

  5. Symbolic regression - learning the symbolic expression of a function based on available data
    • Different methods: genetic programming, Cartesian genetic programming, neural network based
    • Learning functions from physical phenomena: https://space.mit.edu/home/tegmark/aifeynman.html
    • Competition and algorithms: https://cavalab.org/srbench/competition-2022/
    • Dimensionally aware GP - GP that uses information about the dimensionality and units of the data
    • AI Feynman:LINK
    • Linear scaling and interval arithmetic
    • Dealing with interpretability of expressions

  6. Container relocation problem - problem of relocating containers in terminals to board them on a ship
    • General: https://sites.google.com/site/shunjitanaka/brp
    • Optimisation with various heuristics of metaheuristics
    • Various topics:
      • Dynamic problem variant - changes happen during the execution (arrival of new containers, order of retrievel not known)
      • optimising multiple objectives
      • Multibay problems, duplicate containers, possibility to relocate two or mor containers with the crane
      • Optimising energy consumption during operation
    • Stacking in warehouses: https://dynstack.adaptop.at/

  7. Other optimisation problem

  8. Numeric optimisation - implementation of different numeric optimisation methods
    1. Sequential quadratic programming
    2. Memetic algorithms - combination of evolutionary algorithms with numeric optimisation algorithms
    3. Othern numeric optimisation algorithms (gradient based, constraint handling...)

  9. Machine learning - applying machine learning methods to different classification and regression methods