Back to the program.
Hardness prediction for a MMAS on the UCTP
Philipp Kostuch
IRIDIA
Philipp.Kostuch@st-annes.ox.ac.uk

Abstract

I will present a project that has been jointly undertaken with my colleague Krzysztof Socha over the last month. We have attempted to find a statistical model that can predict the performance of a Min-Max Ant system for the University Course Timetabling problem (UCTP). This means, that simply by looking at certain instance characteristics, the objective function value that will be obtained (within a time limit) on this instance by the given algorithm is to be predicted. Firstly, the conceptual framework of this task and its necssary limitations will be introduced. Then, the statistics behind the model, the model selection process and possible applications of this work will be discussed. The statistical side of this work is hoped to be introduced gently and in a non-technical manner.

Keywords

Min-Max Ant system, Timetabling problem (UCTP)