معرفی
Pascal Kerschke is a researcher at TU Dresden, Germany, with extensive contributions to evolutionary computation, continuous optimization, and exploratory landscape analysis. He actively collaborates with leading researchers in automated algorithm selection and metaheuristics.
Research Interests:
- Automated Algorithm Selection using machine learning and landscape features
- Exploratory Landscape Analysis (ELA) for characterizing optimization problems
- Multi-objective and multimodal optimization
- Traveling Salesperson Problem (TSP) and local search heuristics
- Development of benchmark problems and performance indicators
- Integration of deep learning in optimization analysis
Recent Publication Trends: His recent work (2022–2025) emphasizes deep learning for landscape analysis (e.g., Deep-ELA), rigorous benchmarking of optimization algorithms, visualization of multi-objective landscapes, and improving automated algorithm selection through better feature engineering and instance selection. He also investigates theoretical aspects of performance indicators like the R2 indicator.
Scientific Awards: No specific awards are mentioned in the provided data.
Advising and Grants: While no formal grants are listed, his frequent co-authorship with junior researchers such as Lennart Schäpermeier, Moritz Vinzent Seiler, and Jonathan Heins suggests an active mentoring or supervisory role. He contributes to collaborative projects like the Dagstuhl Seminar on benchmarking challenges and works with tools such as OpenML and FLACCO, indicating involvement in community-driven research initiatives.
Labs and Teams: He is part of a vibrant research group at TU Dresden focused on evolutionary algorithms and optimization, collaborating closely with Heike Trautmann, Mike Preuss, and Christian Grimme. His work is integrated into larger efforts in automated machine learning and benchmarking, including participation in workshops and tutorials at major conferences like GECCO and PPSN.
