Papers
16
Total Citations
887
H-Index
13
About
Kristian Kersting is a versatile machine learning researcher whose work spans probabilistic modeling, robotics, agricultural data science, and relational AI. He is perhaps best known for his foundational contribution to heteroscedastic Gaussian process regression, where he pioneered a dual-GP framework capable of modeling input-dependent noise — a paper that has garnered over 300 citations and remains a cornerstone reference in probabilistic machine learning. His expertise in Gaussian processes extends naturally into robotics, where his terrain modeling work enabled legged robots to navigate complex environments through learned, probabilistically rich surface representations. Beyond probabilistic methods, Kersting has made significant contributions to lifted probabilistic inference and relational reinforcement learning, advancing scalable AI reasoning in structured, symmetry-rich domains. His interdisciplinary reach is particularly striking: he has brought machine learning to precision agriculture, co-authoring influential work on hyperspectral sensing for crop disease detection, and explored robotic grasping through probabilistic logic and graph kernel methods. More recently, he has ventured into human-robot interaction, investigating learned facial expression generation for socially intelligent robots. With a citation profile reflecting consistent impact across diverse fields, Kersting exemplifies the modern AI researcher who bridges fundamental theory with compelling real-world application.
Research Focus
Key Achievements
Top Papers
- 1Most likely heteroscedastic Gaussian process regression313 citations · 2007
- 2Lifted Probabilistic Inference105 citations · 2012
- 3
- 4Learning predictive terrain models for legged robot locomotion81 citations · 2008
- 5Data Mining and Pattern Recognition in Agriculture56 citations · 2013
- 6
- 7Exploration in relational domains for model-based reinforcement learning42 citations · 2012
- 8
- 9
- 10