Konstantinos D. Polyzos
University of Minnesota, University of Patras, University of California San Diego
Papers
7
Total Citations
82
H-Index
5
About
Konstantinos D. Polyzos is a researcher at the forefront of Bayesian optimization (BO) and active learning (AL), with a particular focus on enhancing the efficiency and robustness of these methods for real-world applications. His major contributions include pioneering surrogate modeling techniques that move beyond the standard single Gaussian process, as demonstrated in his highly cited 2023 work (53 citations), which significantly improves the performance of BO in complex, expensive-to-evaluate black-box functions. Polyzos has also advanced the field by integrating ensemble learning models with adaptive acquisition functions, such as the adaptive expected improvement, to further optimize sequential decision-making. In active learning, he has developed weighted ensemble strategies that adaptively select the most informative data points for labeling, reducing costs in domains like medical imaging and robotics. His work extends to practical robotics, where he applies Bayesian optimization to view planning for efficient 3D reconstruction in noisy agricultural environments. With a growing citation record and a clear trajectory of innovation, Polyzos is establishing himself as a key contributor to the theory and application of data-efficient machine learning.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3Weighted Ensembles for Active Learning with Adaptivity7 citations · 2022
- 4Solving the Inverse Kinematics of Robotic Arm Using Autoencoders5 citations · 2019
- 5Weighted Ensembles for Adaptive Active Learning5 citations · 2024
- 6
- 7