I compared human recognition of the basic sounds used in the English language with automatic recognition of the same sounds, exploring the kinds of errors made by each to determine their respective strengths and weaknesses. Using knowledge of human speech processing and an understanding of the pattern matching technique used in automatic speech recognition (ASR), I was able to identify potential reasons for the primary weakness in ASR and demonstrate the feasibility of an approach to overcoming it.
I lead a small group of researchers and software developers working on practical applications of probabilistic reasoning using Bayesian networks. This work has the goal of understanding high-level concepts about the environment, such as potential threats and the intentions of other entities. An increasing focus is on using the current understanding to determine what additional information has the most value for clarifying the current situation and how to identify false or erroneous information.
The class work and research I performed while in the SHBT program gave me a solid foundation for my current research. The program's emphasis on core skills such as proposal writing and presenting research to a wide range of audiences continues to benefit me in my daily activities. Most importantly, the experience of merging different fields in the process of performing my research has been particularly useful in my current position, where I regularly have the opportunity to monitor other cutting-edge research efforts in different domains, and to identify the potential benefits of coordinating those technologies with our advances in probabilistic and symbolic reasoning.
