Paper accepted at AAAI'25
The paper uses algorithm engineering to improve a decision tree algorithm.
Published:
Our paper Witty: An Efficient Solver for Computing Minimum-Size Decision TreesExternal link coauthored by Luca Staus, Christian Komusiewicz, Frank Sommer, and Manuel Sorge was selected for presentation at the 39th Annual AAAI Conference on Artificial Intelligence, one of the leading AI conferences.
In the paper, we describe how a theoretical algorithm for the computation of decision trees that make no classification error and have a smallest number of internal nodes can be augmented so that it reaches state-of-the-art running time performance.
The paper will be presented by Luca Staus at the conference.