Daniel Kottke

Papers

Total Citations

H-Index

About

Daniel Kottke is a researcher at the forefront of machine learning and robotics, specializing in active learning and human-in-the-loop systems. His work addresses a critical industrial challenge: enabling robots to autonomously adapt to new tasks—such as object sorting—without expensive manual reprogramming. Kottke’s key contributions include developing probabilistic active learning techniques that allow robots to efficiently query the most informative training examples, dramatically reducing the data needed for adaptation. His 2018 paper, "Active Sorting – An Efficient Training of a Sorting Robot with Active Learning Techniques" (6 citations), demonstrates a practical framework for this approach. Beyond autonomous learning, Kottke tackles the fundamental issue of label quality. In his 2019 work, "Combining Self-reported Confidences from Uncertain Annotators to Improve Label Quality" (4 citations), he pioneered methods to leverage annotators’ self-reported confidence scores, enabling more reliable training data from multiple, potentially uncertain human experts. This dual focus—on efficient robot learning and robust human-in-the-loop annotation—positions Kottke as a key innovator in making intelligent automation both practical and trustworthy for real-world industrial applications.