Automating SAT Solver Research
- Subject:Data-Driven Automatic and Optimal Evaluation of SAT Solvers
Markus Iser, Jakob Bach
Supervised as well as unsupervised learning can be used to exploit the empircally observed complementarity of algorithms for hard combinatorial problems (SAT) which is not well understood yet from an analytical perspective. Explainable machine learning can also provide meaningful insights about the method under evaluation and can moreover help to reduce the number of runtime experiments for evaluating new SAT algorithms. The provided thesis is under co-supervision of Jakob Bach (IPD Böhm, Data Science) and Markus Iser (ITI Sanders, SAT Algorithms).