Kiran Vodrahalli

Columbia University, City University of Seattle

Papers

Total Citations

H-Index

About

Kiran Vodrahalli is a researcher whose work sits at the intersection of machine learning, robotics, and formal logic, with a focus on creating AI systems that are both interpretable and manipulable. His primary contributions center on developing methods for learning policies from expert demonstrations that can be understood and adjusted by humans. In his most-cited work, "Deep Bayesian Nonparametric Learning of Rules and Plans from Demonstrations with a Learned Automaton Prior" (2020, 6 citations), Vodrahalli introduced a novel approach that models high-level action sequences as an automaton, linking them to formal logic to achieve interpretability. This framework allows users to not only see *what* a learned policy does but also to modify its behavior by manipulating the underlying logical rules. He extended this line of research in "Learning and planning with logical automata" (2021, 2 citations), further integrating automata-based representations into planning. While his citation counts are still growing, Vodrahalli’s work is notable for tackling a critical challenge in AI: bridging the gap between powerful black-box models and the need for transparency and human control. His research is particularly relevant for applications in robotics and autonomous systems where safety and trust are paramount.