Multiobjective optimization

PhD IN COMPUTING SEMINAR Prof. Luis Paquete, University of Coimbra, Portugal

When

Sep 13, 2010 10:00 AM to Sep 17, 2010 01:00 PM (Europe/Madrid / UTC200)

Where

Campus Nord UPC, room S208, Omega building.

Add event to calendar

iCal


Dates: September 13, 15 and 17.
Hours: 10:00-13:00
Location: Campus Nord UPC, room S2 208, building Omega.


Sumary:

Multiobjective optimization problems arise in many real-life applications and they are among the hardest optimization problems. Therefore, high-quality approximations that can be obtained in reasonable time are, in practice, preferable to the often infeasible long computation times required for finding the optimum. For this reason, metaheuristic approaches, such as evolutionary algorithms, ant colony optimization and tabu search, have been proposed to solve this class of problems.

The goal of this course is to introduce the students to principles of metaheuristics to solve multiobjective optimization problems, with particular emphasis on combinatorial problems. The course will include introductory topics on multiobjective optimization and advanced topics on design and analysis of metaheuristics for this class of problems. Several applications and software tools will be discussed during the course. After the course, the student is expected to: i) Adapt metaheuristic principles and design new metaheuristics to new optimization problems and ii) analyse the performance of metaheuristics with respect to solution quality, computation time and memory usage.

The lectures will include the following topics:

  1. Introduction to Multiobjective Combinatorial Optimization
    • Notions of optimality
    • Complexity and intractability
    • Solution Methods
  2. Multiobjective Metaheuristics
    • Single-objective Metaheuristics
    • Component-wise Methods
    • Indicator-based Methods
    • Scalarized Methods
    • Multiobjective Metaheuristics
  3. Performance Assessment and Software frameworks
    • Experimental analysis of Metaheuristics
    • Unary Indicators
    • Binary Indicators
    • Attainment Functions
    • Software Frameworks

Notes:


The course is open to any interested person

For PhD or Master students it is possible to get additional ECTS credits for this course, please contact Mercè Juan(merce@lsi.upc.edu) indicating your situation.

The course is organized under the mobility program for professors in PhD Programs awarded with the quality mention by the Spanish Ministry of Education.

Keywords
Seminar