(AGENPARL) - Roma, 6 Giugno 2023 - (AGENPARL) – VENEZIA mar 06 giugno 2023 Speakers: 
Federica Sarro, UCL – University College London, UK
Rebecca Moussa, UCL – University College London, UK

Abstract:
In this talk we will introduce a novel approach based on multi-objective evolutionary search to automatically generate optimal ensemble classifiers. We dubbed this approach as Multi-objective Ensemble Generation (MEG) and assessed its effectiveness for software defect prediction (DP) by empirically benchmarking it with respect to state-of-the-art ensembles. Our results show that MEG can generate ensembles which produce similar or more accurate predictions than those achieved by all the other approaches considered in 73% of the cases (with favourable large effect sizes in 80% of them). MEG is not only able to generate ensembles that yield accurate defect predictions, but it also does it automatically. While we evaluated the effectiveness of MEG for Software Defect Prediction, MEG can be applied to other problems, and we invite both researchers and practitioners to further explore its effectiveness for DP as well as other domains. To this end, we have made MEG source code publicly available.

This talk is based on the work by Rebecca Moussa, Giovani Guizzo, and Federica Sarro. 2022. MEG: Multi-objective Ensemble Generation for Software Defect Prediction. In Procs. of the 16th ACM/IEEE International Symposium on Empirical Software Engineering and Measurement. DOI: 10.1145/..

Bio Sketch:
Federica Sarro is a Professor of Software Engineering at University College London, where she is the Head of the Software System Engineering group and has established and lead the SOLAR group within the CREST centre. Her research covers Predictive Analytics for Software Engineering, Empirical Software Engineering and Search-Based Software Engineering, with a focus on automated software management, optimisation, testing and repair. On these topics she has published over 80 peer-reviewed scholarly articles in well-renowned Software Engineering venues. She has also received several international awards, latest the IEEE Technical Council Software Engineering Rising Star Award for her excellence in Software Engineering research with scholarly and real-world impact.

Rebecca Moussa is a PhD candidate and Research Assistant at at University College London advised by Profs. Federica Sarro and Mark Harman. At UCL Rebecca is member of SSE, SOLAR and CREST. Her main research interest is Software Analytics, and her work has contributed to this field both with technological and methodological advances.

Web page: https://solar.cs.ucl.ac.uk/

Fonte/Source: http://www.unive.it/data/agenda/1/75731