Home

Several formal models for evaluating non-discrimination in machine learning algorithms resort to formulations using propositional logic. The formulations that result from these models have been getting bigger and bigger, so finding optimal solutions for these formulations is increasingly difficult. The alternative is to find solutions as close as possible to the optimal value. This scholarship intends to develop new approximation algorithms for the MaxSAT problem. New ways of partitioning soft clauses in MaxSAT will be developed. After obtaining the partitions, we will use multi-objective optimization algorithms to optimize each of the partitions and obtain a better approximation to the MaxSAT problem.