Matthias Poloczek

Papers

Total Citations

154

H-Index

About

Matthias Poloczek is a prominent researcher specializing in Bayesian optimization and machine learning, with a particular focus on developing scalable and efficient methods for optimizing expensive black-box functions. His work addresses some of the most pressing challenges in the field, including extending Bayesian optimization to high-dimensional problem spaces and large-scale datasets that have historically resisted such approaches. Poloczek's most influential contribution, "Scalable Global Optimization via Local Bayesian Optimization" (2019), has garnered 144 citations and represents a significant breakthrough in making Bayesian optimization practical for real-world applications involving thousands of observations and complex, high-dimensional landscapes. By leveraging local optimization strategies within a global framework, his work dramatically expanded the practical utility of these methods. His subsequent research, "Increasing the Scope as You Learn: Adaptive Bayesian Optimization in Nested Subspaces" (2023), continues this trajectory by developing adaptive techniques that progressively explore higher-dimensional spaces, with promising applications in life sciences, neural architecture search, and robotics. Through these contributions, Poloczek has helped transform Bayesian optimization from a theoretically elegant but practically limited tool into a robust methodology capable of tackling impactful, large-scale scientific and engineering challenges.