Overview of Work
My current research efforts are focused in the fields of feature engineering, feature learning, and machine learning for multivariate, time-series data sets. In particular, my research hopes to develop a methodology for automatically detecting a state-space without prior knowledge of the underlying system. This objective is currently being evaluated through the analysis of both individual data signals as well as how these signals relate to one another over time. Once a preliminary state space is developed, an understanding of probabilities for moving into each state space can be evaluated. In addition, given the automated nature of the state space development, new states can be discovered in real-time. Both the preliminary state space as well as newly discovered states can be evaluated after discovery by a subject-matter expert in order to justify decisions.
Current projects include:
- Detection and prediction of hypoxic states in fighter pilots
- Detection and prediction of attention-related human performance limiting states
- Building infrastructure state analysis
Previous projects include:
- Validation of crash events in fleet vehicles