Evolutionary Computation I (ECI)

CHAIR: ANDREJ DOBNIKAR

Monday, March 21st, 11h00-12h40

Paper ID   Title
   
ECI-1 Evolutionary Design and Evaluation of Modeling System for Forecasting Urban Airborne Maximum Pollutant Concentrations
ECI-2 Evolving Evolvability: Evolving both representations and operators
ECI-3 A Multi-Objective Evolutionary Algorithm for Solving Traveling Salesman Problems: Application to the Design of Polymer Extruders
ECI-4 The Pareto-Box Problem for the Modelling of Evolutionary Multiobjective Optimization Algorithms


ECI-1
 
Title: Evolutionary Design and Evaluation of Modeling System for Forecasting Urban Airborne Maximum Pollutant Concentrations
Author(s): H. Niska,
T. Hiltunen,
A. Karppinen,
M. Kolehmainen
Abstract: In this paper, a modeling system based on the combination of a multi-layer perceptron (MLP), a meteorological pre-processing model and a numerical weather prediction model (NWP) is developed and evaluated for the forecasting of urban airborne maximum pollutant concentrations. As an important phase of the system design, the multi-objective genetic algorithm (MOGA) and the sensitivity analysis of MLP are used in combination for identifying feasible system inputs. The final evaluation of the modeling system is performed by utilizing the hourly concentrations of nitrogen dioxide (NO2), particulate matter (PM10), fine particulate matter (PM2.5) and ozone (O3) measured at an urban air quality station in central Helsinki (capital of Finland) during the period from 1.5.2000 to 1.5.2003. The study showed that the combination of MOGA and the sensitivity analysis is an appropriate tool for selecting inputs of neural network and can be recommended for wider dissemination and use. The results showed good general performance of the modeling system. However, capability to model episodic conditions was only moderate.


ECI-2
 
Title: Evolving Evolvability: Evolving both representations and operators
Author(s): Grant W. Braught
Abstract: The behavior of an evolutionary system incorporating both an evolving genetic representation (a learning mechanism) and an evolving genetic operator (mutation) is explored. Simulations are used to investigate the co-adaptation of these two self-adaptive mechanisms. The results illustrate a duality between these two mechanisms in their production of a transmission function. Further, the adaptive power of the representation is shown to affect the balance in this duality and to influence the conservatism of the transmission function.


ECI-3
 
Title: A Multi-Objective Evolutionary Algorithm for Solving Traveling Salesman Problems: Application to the Design of Polymer Extruders
Author(s): A. Gaspar-Cunha
Abstract: A Multi-Objective Evolutionary Algorithm (MOEA) for solving Traveling Salesman Problems (TSP) was developed ande used in the design of screws for twin screw polymer extrusion. This is an important and original contribution in the design of these machines. The Twin-Screw Configuration Problem (TSCP) can be formulated as a TSP A different MOEA is developed, in order to take into account the discrete nature of the TSCP. The algorithm proposed was applied to some case studies where the practical usefulness of this approach was demonstrated. Finally, the computational results are confronted with experimental data showing the validity of the approach proposed.


ECI-4
 
Title: The Pareto-Box Problem for the Modelling of Evolutionary Multiobjective Optimization Algorithms
Author(s): Mario Koeppen,
Raul Vicente-Garcia,
Bertram Nickolay
Abstract: This paper presents the Pareto-Box problem for modelling evolutionary multi-objective search. The problem is to find the Pareto set of randomly selected points in the unit hypercube. While the Pareto set itself is only comprised of the point 0, this problem allows for a complete analysis of random search and demonstrates the fact that with increasing number of objectives, the probability of finding a dominated vector is decreasing exponentially. Since most nowadays evolutionary multi-objective optimization algorithms rely on the existence of dominated individuals, they show poor performance on this problem. However, the fuzzification of the Pareto-dominance is an example for an approach that does not need dominated individuals, thus it is able to solve the Pareto-Box problem even for a higher number of objectives.