Detail publikace

Heuristic challenges for spatially distributed waste production identification problems

Originální název

Heuristic challenges for spatially distributed waste production identification problems

Anglický název

Heuristic challenges for spatially distributed waste production identification problems

Jazyk

en

Originální abstrakt

The aim of the paper is to present the advances in the development of optimization foundations of software tool Justine designed for the forecasting of spatially distributed waste production under incomplete, uncertain, and even contradictory information. Justine tool has been already successfully used for practical computations that serve for investment planning in the area of waste processing unit allocation and design. However, the experience with real-world oriented computations generates new modeling and algorithmic challenges linked to the future use of this tool. Specifically, the obtained data are related to the existing structure of regions and their subregions, and hence, because of various demographical, geographical, and industry related reasons, the data are often of various quality and heterogeneous nature from the quantitative point of view. Therefore, the computational model developed for Justine tool is modified to deal with possibility to reorganize the process of collecting and clustering data in such a way that a suitable regression-based criterion is minimized. Because of the presence of binary variables and the fact that their number is extremely large for the real-world data we suggest to implement a suitable heuristic. The paper introduces the first step in this direction and states a challenge to include more advanced heuristics in the future.

Anglický abstrakt

The aim of the paper is to present the advances in the development of optimization foundations of software tool Justine designed for the forecasting of spatially distributed waste production under incomplete, uncertain, and even contradictory information. Justine tool has been already successfully used for practical computations that serve for investment planning in the area of waste processing unit allocation and design. However, the experience with real-world oriented computations generates new modeling and algorithmic challenges linked to the future use of this tool. Specifically, the obtained data are related to the existing structure of regions and their subregions, and hence, because of various demographical, geographical, and industry related reasons, the data are often of various quality and heterogeneous nature from the quantitative point of view. Therefore, the computational model developed for Justine tool is modified to deal with possibility to reorganize the process of collecting and clustering data in such a way that a suitable regression-based criterion is minimized. Because of the presence of binary variables and the fact that their number is extremely large for the real-world data we suggest to implement a suitable heuristic. The paper introduces the first step in this direction and states a challenge to include more advanced heuristics in the future.

BibTex


@article{BUT128210,
  author="Vlastimír {Nevrlý} and Radovan {Šomplák} and Pavel {Popela} and Martin {Pavlas} and Ondřej {Osička} and Jakub {Kůdela}",
  title="Heuristic challenges for spatially distributed waste production identification problems",
  annote="The aim of the paper is to present the advances in the development of optimization foundations
of software tool Justine designed for the forecasting of spatially distributed waste production under incomplete, uncertain, and even contradictory information. Justine tool has been already successfully used for practical computations that serve for investment planning in the area of waste processing unit allocation and design. However, the experience with real-world oriented computations generates new modeling and algorithmic challenges linked to the future use of this tool. Specifically, the obtained data are related to the existing structure of regions and their subregions, and hence, because of various demographical, geographical, and industry related reasons, the data are often of various quality and heterogeneous nature from the quantitative point of view. Therefore, the computational model developed for Justine tool is modified to deal with possibility to reorganize the process of collecting and clustering data in such a way that a suitable regression-based criterion is minimized. Because of the presence of binary variables and the fact that their number is extremely large for the real-world data we suggest to implement a suitable heuristic. The paper introduces the first step in this direction and states a
challenge to include more advanced heuristics in the future.",
  address="VUT",
  chapter="128210",
  howpublished="print",
  institution="VUT",
  number="1",
  volume="2016",
  year="2016",
  month="june",
  pages="109--116",
  publisher="VUT",
  type="journal article in Scopus"
}