Finale Doshi

Massachusetts Institute of Technology, University of Cambridge

Papers

Total Citations

126

H-Index

About

Finale Doshi is a leading researcher in human-robot interaction and spoken dialogue systems, with a focus on developing robust, adaptive agents that can navigate the inherent uncertainty of natural language communication. Her key contributions lie in applying probabilistic models, particularly Partially Observable Markov Decision Processes (POMDPs), to dialog management and human-robot interaction. In her highly cited 2007 work (67 citations), Doshi pioneered efficient model learning for dialog management, demonstrating how intelligent planning algorithms can be made robust to the ambiguity of human speech. She extended this framework in her 2008 paper (59 citations) by creating an adaptive human-robot interaction system that explicitly models and manages uncertainty during spoken language exchanges, allowing robots to recover from communication errors and better understand user intentions. This work has been foundational for developing more resilient, user-friendly conversational agents. Doshi’s research bridges theoretical advances in probabilistic planning with practical, real-world applications, making her a notable figure in the fields of adaptive dialogue systems and socially intelligent robotics.