M.Sc. Michel Medema
PhD Student
- room number: 566
- e-mail: m.medema [at] rug.nl
- personal website: N/A
Research
- Discrete Optimisation Problems
- Distributed Systems
- Big Data
Recent publications
- Correlating the Community Structure of Constraint Satisfaction Problems with Search Time (), In International Journal on Artificial Intelligence Tools, 2022.
Abstract
A constraint satisfaction problem (CSP) is, in its most general form, an NP-complete problem. One of the several classes of tractable problems that exist contains all the problems with a restricted structure of the constraint scopes. This paper studies community structure, a particular type of restricted structure underpinning a class of tractable SAT problems with potentially similar relevance to CSPs. Using the modularity, it explores the community structure of a wide variety of problems with both academic and industrial relevance. Its impact on the search times of several general solvers, as well as one that uses tree-decomposition, is also analysed to determine whether constraint solvers exploit this type of structure. Nearly all CSP instances have a strong community structure, and those belonging to the same class have comparable modularity values. For the general solvers, strong correlations between the community structure and the search times are not apparent. A more definite correlation exists between the modularity and the search times of the tree-decomposition, suggesting that it might, in part, be able to take advantage of the community structure. However, combined with the relatively strong correlation between the modularity and the tree-width, it could also indicate a similarity between these two measures.
BibTeX
urldoi - Automated Service Composition Using AI Planning and Beyond (), Chapter in (M. Aiello, A. Bouguettaya, D. A. Tamburri, W. J. van den Heuvel, eds.), Springer International Publishing, 2021.
- Energy Consumption Patterns and Load Forecasting with Profiled CNN-LSTM Networks (), In Processes, volume 9, 2021.
Abstract
By virtue of the steady societal shift to the use of smart technologies built on the increasingly popular smart grid framework, we have noticed an increase in the need to analyze household electricity consumption at the individual level. In order to work efficiently, these technologies rely on load forecasting to optimize operations that are related to energy consumption (such as household appliance scheduling). This paper proposes a novel load forecasting method that utilizes a clustering step prior to the forecasting step to group together days that exhibit similar energy consumption patterns. Following that, we attempt to classify new days into pre-generated clusters by making use of the available context information (day of the week, month, predicted weather). Finally, using available historical data (with regard to energy consumption) alongside meteorological and temporal variables, we train a CNN-LSTM model on a per-cluster basis that specializes in forecasting based on the energy profiles present within each cluster. This method leads to improvements in forecasting performance (upwards of a 10% increase in mean absolute percentage error scores) and provides us with the added benefit of being able to easily highlight and extract information that allows us to identify which external variables have an effect on the energy consumption of any individual household.
BibTeX
urldoi - A multi-robot allocation model for multi-object based on Global Optimal Evaluation of Revenue (), In International Journal of Advanced Robotic Systems, volume 18, 2021.
Abstract
The problem of global optimal evaluation for multi-robot allocation has gained attention constantly, especially in a multi-objective environment, but most algorithms based on swarm intelligence are difficult to give a convergent result. For solving the problem, we established a Global Optimal Evaluation of Revenue method of multi-robot for multi-tasks based on the real textile combing production workshop, consumption, and different task characteristics of mobile robots. The Global Optimal Evaluation of Revenue method could traversal calculates the profit of each robot corresponding to different tasks with global traversal over a finite set, then an optimization result can be converged to the global optimal value avoiding the problem that individual optimization easy to fall into local optimal results. In the numerical simulation, for fixed set of multi-object and multi-task, we used different numbers of robots allocation operation. We then compared with other methods: Hungarian, the auction method, and the method based on game theory. The results showed that Global Optimal Evaluation of Revenue reduced the number of robots used by at least 17%, and the delay time could be reduced by at least 16.23%.
BibTeX
urldoi - The Community Structure of Constraint Satisfaction Problems and Its Correlation with Search Time (), In 2020 IEEE 32nd International Conference on Tools with Artificial Intelligence (ICTAI), 2020.
(For more publications go to Michel's publication page)