Msc thesis in collaboration with Bordeaux University - Benchmarking, SAT solving, Automated algorithm configuration
34 views
Skip to first unread message
marie.a...@gmail.com
Feb 10, 2021, 5:06:13 AM2/10/21
to LIACS thesis projects
This project aims at developing an approach to efficiently compare the expected performance of two algorithms by carefully selecting a few problem instances on which to run the new method.
It can be applied to newly developed methods to solve computationally hard problems, particularly NP-hard problems, that needs to be formally compared with existing state-of-the-art methods, leading to extensive, time-consuming and resource-intensive experimentations.
Such an approach would also benefit automated algorithm configuration whose purpose is to automate the choice of (hyper-)parameter values for a given algorithm. Deciding which among two possible sets of parameter values is the best is crucial for such configuration methods.
It can be applied to newly developed methods to solve computationally hard problems, particularly NP-hard problems, that needs to be formally compared with existing state-of-the-art methods, leading to extensive, time-consuming and resource-intensive experimentations.
Such an approach would also benefit automated algorithm configuration whose purpose is to automate the choice of (hyper-)parameter values for a given algorithm. Deciding which among two possible sets of parameter values is the best is crucial for such configuration methods.
Reply all
Reply to author
Forward
0 new messages