| On 2022-11-08 at 11:15:00 (Brussels Time) |
Abstract
Algorithm configuration is the task of finding a set of parameter settings which allows an algorithm to reach an expected performance level. All sorts of algorithms are commonly designed to expose a set of parameters for its later configuration. Such a decision is made given that algorithms are often required to be flexible and general enough, and parameters enable them to be adapted when applied. Finding adequate settings often requires expert knowledge of both the algorithm to be configured and the application context at hand. For a human, the process can be tedious, and from a practical perspective, it is also computationally expensive. Since the configuration is the last step performed to adjust an algorithm for its application, the task is very relevant in different domains such as optimization, machine learning and software development. In this talk, we will discuss algorithm configuration, how this task can be automated and what are the possibilities that this automation offers. Short Bio of the Speaker
Leslie Pérez Cáceres is a professor at Pontificia Universidad Católica de Valparaíso, Chile since 2018. She is also the Director of the Artificial Intelligence Diploma of the PUCV’s Escuela de Ingeniería Informática. She received the M.S. degree in Engineering Sciences in 2011 from the Universidad Técnica Federico Santa María and, the Ph.D. in Engineering and Technology Sciences from the Université Libre de Bruxelles in 2017. Her research interests are the automatic configuration of optimization algorithms and the design of optimization algorithm for solving combinatorial optimization problems. She is also one of the developers of the irace configuration tool.
