The term Pervasive Informatics refers to harvesting and interpreting information from cyber physical boundary environments for decision support. To do this we pursue fundamental advancements in string processing and information retrieval in terms of performance and robustness. We also apply the state of the art in machine learning and statistical processing to infer semantics where previously they could not be logically assigned. In addition, we assume the cyber physical boundary to be a lossy environment, meaning that noise and incompleteness becomes a major factor in pattern recognition so approximate matching and probabilistic reasoning is vital.
However, solutions using these tools are not immediately applicable from problem to problem, so we put much effort into tailoring algorithms, preprocessing data and extracting features before undertaking modeling and matching tasks. Further, we apply ourselves to relevant, real world problems (medical, security, mission awareness and illicit activities). These problems not only push the cyber-physical boundary, but also computational boundaries in terms of large scale/high speed data processing and the curse of dimensionality.
Research Team
- Hao Wang
- John Pecarina
- Mike George
- Christopher Bodden