This shows you the differences between two versions of the page.
| Both sides previous revision Previous revision | Next revision Both sides next revision | ||
|
teaching:mfe:ia [2013/04/03 10:56] stuetzle [Stochastic local search algorithms for weighted maximum clique problems.] |
teaching:mfe:ia [2013/04/03 11:06] stuetzle |
||
|---|---|---|---|
| Line 248: | Line 248: | ||
| * [[http://iridia.ulb.ac.be/~fmascia|Franco Mascia (IRIDIA)]] | * [[http://iridia.ulb.ac.be/~fmascia|Franco Mascia (IRIDIA)]] | ||
| + | |||
| + | ===== Feature Extraction and Automatic Algorithm Selection. ====== | ||
| + | |||
| + | The performance of (Stochastic Local Search) algorithms for a given problem depends on the algorithm design and on the setting of the algorithm's parameter. Given a heterogeneous set of instances for a given problem a good algorithm design (or parameter configuration) for one instance is not necessary the best design for all instances. On the contrary a tuning of an algorithm on a specific family of similar instances may affect negatively its performance on other families of instances. | ||
| + | |||
| + | The thesis will focus on devising automatic methods for extracting features from the instances, select the relevant features, and learning (in the framework of multi-class classification) the | ||
| + | relationship, if there is one, between the instances features and the best algorithm for the instance. The results will be instrumental for algorithm selection or the creation of portfolios of complementary algorithms suitable for large sets of diverse instances for a given problem. | ||
| + | |||
| + | |||
| + | Required skills: good knowledge of C or C++ programming and of a scripting language (e.g., python); good knowledge of machine learning methods would also be helpful. | ||
| + | |||
| + | |||
| + | * Contacts : | ||
| + | * [[http://iridia.ulb.ac.be/~stuetzle|Thomas Stützle (IRIDIA)]] | ||
| + | * [[http://iridia.ulb.ac.be/~fmascia|Franco Mascia (IRIDIA)]] | ||