Hasso-Plattner-Institut
Prof. Dr. Tobias Friedrich
  
 

15.01.2020

Two papers accepted at ECAI and EvoCOP

Europe’s premier AI Research venue, the European Conference on Artificial Intelligence (ECAI), holds its 24th edition in June in Santiago de Compostela, Spain. The Algorithm Engineering group contributes one paper. This work suggests that a simple greedy algorithm can be used to maximize non-monotone submodular functions under complex knapsack constraints. This problem finds several applications in AI and machine learning.

In addition to that, the 20th European Conference on Evolutionary Computation in Combinatorial Optimisation (EvoCOP) accepted another paper of the group. This paper disproves a previous claim that a certain type of evolutionary algorithm is not well suited for optimizing a deceptive landscape. In fact, the considered algorithm is even more efficient than any competing algorithm deemed efficient for that setting. Due to this surprising insight, the paper is nominated for a best-paper award.

  • Non-Monotone Submodular M... - Download
    Doskoč, Vanja; Friedrich, Tobias; Göbel, Andreas; Neumann, Aneta; Neumann, Frank; Quinzan, FrancescoNon-Monotone Submodular Maximization with Multiple Knapsacks in Static and Dynamic Settings. European Conference on Artificial Intelligence (ECAI) 2020
     
  • The Univariate Marginal D... - Download
    Doerr, Benjamin; Krejca, Martin S.The Univariate Marginal Distribution Algorithm Copes Well With Deception and Epistasis. Evolutionary Computation in Combinatorial Optimisation (EvoCOP) 2020