The Dean of the Faculty of Electrical Engineering of the Czech Technical University in Prague awarded the Dean’s Prize for a prestigious dissertation to three researchers in 2023 including Erik Derner from CIIRC CTU.
The following researchers received the award:
Ing. Tomáš Dlask for his thesis Block-Coordinate Descent and Local Consistencies in Linear Programming under the supervision of his supervisor doc. Ing. Tomáš Werner, Ph.D., supervisor.
Ing. David Futschik for the thesis Leveraging machine learning for artistic stylization under the supervision of the supervisor prof. Ing. Daniel Sýkora, Ph.D.
Ing. Erik Derner for the work Data-efficient methods for model learning and control in robotics under the supervision of the supervisor prof. Dr. Ing. Robert Babuska.
Erik Derner received his Ph.D. in Control Engineering and Robotics from the Czech Technical University (CTU) in Prague in 2022. In 2023, he joined the ELLIS Alicante Unit as an ELLIS postdoctoral researcher in the team of Dr. Nuria Oliver. His research focuses on the societal and ethical aspects of human-AI interaction, particularly addressing biases in large language models for underrepresented languages. Within the ELLIS Postdoc program, he collaborates with the team of Prof. Robert Babuška at the Czech Institute of Informatics, Robotics, and Cybernetics on safe and robust robotic systems leveraging generative AI. Furthermore, he studies the security aspects and safeguards of large language models. His areas of interest comprise human-centric AI, natural language processing, robotics, computer vision, reinforcement learning, and genetic algorithms.
The thesis „Data-Efficient Methods for Model Learning and Control in Robotics,“ supervised by Prof. Robert Babuška, was developed within the project Robotics for Industry 4.0. It addresses the challenges of data-driven model learning in robotics. It focuses on building models of robots and their environments from a small amount of data, in contrast with the popular and broadly used data-intensive deep neural networks. It introduces several variants and extensions of symbolic regression for robot model learning and control. It also presents a novel sample-selection method to build informative subsets of large data collections. Furthermore, it introduces a change detection method for improved localization of mobile robots in dynamic environments. The proposed methods have been extensively evaluated in real-world scenarios. The thesis comprises publications in top-tier journals and conferences, garnering significant recognition and citations within the robotics community. It has received favorable reviews from opponents from leading European academic institutions.
Source: FEE CTU
