|
Invited Session on Automated Algorithm Design for Multi-objective Optimization Problems
at the 25th International Conference on Multiple Criteria Decision Making (MCDM 2019), Istanbul/ Turkey, June 16 - 21, 2019
Abstract submission deadline: January 30, 2019
Submission: https://mcdm2019.org
Call for Abstracts (Max. 4000 characters with spaces):
--------------------------------------------------------------------
# Subject Area Code: IS05 (Cai, Misir, Ozcan)
Algorithm design is a challenging task, in particular for solving computationally hard optimization problems, which would require expertise both on algorithm development and target problem domain. Due to the presence of immensely large design spaces, the design challenge becomes harder to deal with and time consuming no matter what the expertise level is. Besides that, the designed algorithms are doomed to be sub-optimal so while an algorithm performs well on a certain set of instances from a problem domain, it performs poorly on others. One way of addressing this issue is to design meta-algorithms, under Automated Algorithm Design (a.k.a. computer-aided algorithm design), for delivering high-level algorithmic solutions through selection, configuration and generation operating either Offline, i.e. prior to an instance is being solved, or Online, i.e. while an instance is being solved. Various automated algorithm design approaches have been successfully applied to a wide range of combinatorial optimization problems from academia and real-world, such as timetabling, scheduling, rostering, routing, cutting and packing.
In terms of selection (Algorithm Selection), the idea is to specify the best algorithm(s) for a given problem (~instance) so that the strengths of different algorithms have been combined rather than relying on a single algorithm. Regarding configuration (Algorithm Configuration), the goal is tuning or adaptively/dynamically controlling the parameters of an algorithm while generation (Algorithm Generation) is all about automatically producing algorithms from scratch based on predefined components.
The aim of this special session is to gather researchers and practitioners working on the automated algorithm design issues for Multi-objective Optimization to share their studies, experience and knowledge. The main topics of interest include (but are not limited to):
- hyper-heuristics for multi-objective optimization
- adaptive operator selection for multi-objective optimization
- parameter tuning/control for multi-objective algorithms
- algorithm/operator generation for multi-objective optimization
This special session will be organized in connection with the Task Force on Hyper-heuristics within the Technical Committee of Intelligent Systems and Applications at IEEE Computational Intelligence Society and the EURO working group on Data Science meets Optimization (DSO).
Organizers:
---------------
- Dr. Xinye Cai, Nanjing University of Aeronautics and Astronautics, China
- Dr. Mustafa Misir, Nanjing University of Aeronautics and Astronautics, China
- Dr. Ender Ozcan, University of Nottingham, UK
Important Dates:
-----------------------
Abstract Submissions (Extended): February 27, 2019
Notification to Authors: March 20, 2019
Early Registration: April 10, 2019
Late Registration: May 8, 2019
Conference: June 16 - 21, 2019
|
