Paper Info
Title
Combining parallel coordinates with multi-objective evolutionary algorithms in a real-world optimisation problem
Abstract
Optimisation problems based upon real-world instances often contain many objectives. Many existing Multi-Objective Evolutionary Algorithm techniques return a set of solutions from which the user must make a final selection; typically such a set of solutions may take the form of a non-dominated set. The size of such fronts, especially for larger numbers of objectives, can make it difficult for the user to make a selection of the final solution. This paper outlines an initial investigation into combining elements of Parallel Coordinate plots with multi-objective evolutionary algorithms to allow the user to specify solution areas of interest prior to executing the algorithm. The algorithm encourages the evolution of solutions in these areas through selection pressure. The user is presented with one solution from each area on a Parallel Coordinates plot allowing a simple, informed decision as to the solution to be chosen. This paper uses a Workforce Scheduling and Routing Problem (WSRP) to demonstrate the approach. The WSRP formulation used was previously cited in literature as a multi-objective problem, we formulate it as a 5 objective problem. Our initial results suggest that this approach has potential and is worth investigating further.
Year
DOI
Venue
2017
10.1145/3067695.3082485
GECCO (Companion)
Keywords
Field
DocType
Evolutionary Algorithms, Transportation, Multi-Objective Optimisation, Real-World Problems
Workforce scheduling,Mathematical optimization,Evolutionary algorithm,Computer science,Parallel coordinates,Artificial intelligence,Machine learning
Conference
ISBN
Citations 
PageRank 
978-1-4503-4939-0
0
0.34
References 
Authors
8
1
Name
Order
Citations
PageRank
neil b urquhart18314.70