CIRCA lunchtime seminar

Our speakers are:

Carla Biermann
Title: Sampling solutions to constraint satisfaction problems
Abstract: State-of-the-art constraint programming solvers have become efficient in finding one solution to a specified problem. However, sometimes, users would like to choose a solution from multiple or inspect a sample of solutions without obtaining all solutions, especially when the solution space is large. This need gives rise to solution sampling, which aims to provide the user with an interesting sample of solutions. There are different measures of “interestingness” and algorithms that ensure certain sampling behaviours. I will present two such algorithms and the results of an experimental analysis of their performance on specific constraint satisfactory problems (CSPs). The findings aim to spark a discussion with the audience on more applications of solution sampling in CSPs for my fourth-year project.

and

Nguyen Dang
Title: Automated algorithm configuration and algorithm selection
Abstract: Many algorithms have their own parameters that can impact the algorithm performance. Automated algorithm configuration is a family of general-purpose techniques that focuses on finding the best parameter setting of an algorithm for a given problem. Automated algorithm selection, on the other hand, is a closely related topic that helps us select the best algorithm among a given set of algorithms for a specific instance of a problem. Automated algorithm configuration and algorithm selection have got several successful applications across different domains. In this talk, I will first give a brief introduction to the two topics. The remaining part of the talk will be an open discussion with the audience to see if there are potential applications of those techniques on the GAP system.